Experience Mapping Problem – Case Study: SUPERHOT VR

July 30, 2021 Leave a comment

For an introduction to the Experience Mapping Problem and the goal of hyperrealism in VR, see this post.

SUPERHOT VR succeeds in creating an intoxicating player power fantasy. Time moves at your command, allowing you to execute action moves worthy of a scene in the Matrix. It delivers this powerful fantasy while maintaining a highly abstract graphical style. This game is as far from “photoreal” as you can get. It is also very immersive — people get wholly absorbed by the game:


(video source)

SUPERHOT quickly builds a simple and understandable mapping from life to game, using verbs like “grab” and “shoot”. To analyze how it achieves this rapid, simple readability, we need to characterize the way we unconsciously categorize the world around us.

Suppose I am looking at a shelf full of objects. If I am moving about the room normally, the shelf exists as a single thing to me, a symbol tagged “shelf”. Under normal circumstances, all my knowledge and perception about this collection of molecules is filed under “shelf”.

However, if I become interested in a particular object on the shelf, the symbol of “shelf” decomposes into a slew of new symbols. The object of interest is a symbol (say, “box”). I may also have symbols like “figurine” and “row of books” for the other things on the shelf. The “shelf” symbol still exists, but there is a Symbol Hierarchy now. “Box” and “figurine” have a relation to “shelf” (they are on it), but they also have their own properties.

If I am curious about the contents of the box, the “box” symbol decomposes into “box” and “contents”. (Note that I use the same word (“box”) for both the box-and-contents as well as the box-itself. This word→symbol overlap is a common source of confusion during any discourse, and we must stay wary of it.)

If I want to open the box to see what the contents are, the “box” symbol decomposes into “box” and “box lid”.

Note that Symbol Hierarchies tend to decompose based on the verbs in the context. If the context includes verbs like “open”, then the box symbol hierarchy decomposes into box and box lid. If the context has the verb “grab”, suddenly the shelf decomposes into grabbable objects (“figurine”, “book”, “box”). And if I discover that the figurine is glued to the shelf and therefore cannot be picked up, its symbol is somewhat absorbed by its parent: the figurine object becomes part of the “shelf” symbol again, in the context of “grabbing” or “picking up”; any relationship you have with the shelf (e.g. too heavy to pick up) is shared with the figurine too, and if you knock the shelf down, the figurine will also fall to the ground.

Within this framework, it is obvious why SUPERHOT can deliver an effective, immersive fantasy. The first thing the game does is introduce a verb — “grab” — and teach that all black objects can be grabbed. Second, when you are holding an object that looks like a gun, you can “shoot” the gun. Third, there are red enemies. Red enemies die if a black thing touches them. Then there are other rules, like time moves when you move, you die if a bullet touches you, and white objects are inert and will stop black objects.

The set of symbols and verbs is now defined for the game, and there is a 1-to-1 correspondence between sensory perception and symbol. A gun is always a “gun”. It never decomposes into “slide”, “magazine”, or “grip”. White objects are always “the background”. Even if it looks like a console with buttons and a telephone handset, the player will never even try to decompose the “background” symbol into “buttons” or “telephone”. There are no properties that belong to some white objects but not others; they are always immovable, and impervious to bullets.

This very simple model means that the player rarely encounters a mismatch between expectation and outcome. If I throw a gun, it kills enemies, just like any other thrown black object. I can take cover behind a small flimsy chair, since it is white. I can kill an enemy with a thrown ashtray, since it is black. The player never encounters a mismatch because they never attempt to improperly decompose a Symbol Hierarchy. The game clearly indicates whether something is atomically a “gun”, a “black object”, an “enemy”, or “the background” and never requires you to further divide those atomic symbols.

The player is free to map additional narrative and emotional properties onto these symbols. “Enemies” become an attractor for all the qualities of antagonists from movies and TV, and the suggestive background environments help encourage this emotive transfer: airport, shopping mall, rooftop. These are contexts we have seen action scenes occur. Similarly, the guns and ninja stars are symbolic attractors and gain all the forbidden, sexy power that weapons are granted in other media contexts.

Finally, the game allows the player to perform kinesthetically pleasing sequences with these verbs and symbols. Sure, the enemies just burst into triangles when “killed”, and the guns are simplistic black blobs. Yes, “shooting an enemy” means tiny black blobs came out of your little black blob and hit the big red blob. But you just “punched” an “enemy”, “grabbed” his “gun”, and “shot” another “enemy”! The interaction of symbols carries all the weight the player has put behind the symbols, and the fact that “punching” is kinesthetically similar enough to a real punch (same with “ducking” and “aiming”) allows the player to map back from symbol to reality.

The important take-away is this: because the player quickly builds an accurate model of this virtual world and NEVER encounters a mismatch between expectation and outcome, they stay immersed. And because the objects behind the symbols enable generic symbol assignment (the enemies are not “storm troopers”, they are just archetypal “enemies”), they attract lots of affective qualities. An immersive virtual world filled with high-affect symbols is, well, hyper-real.

Filed under Analysis, Blog, Games, VR Tagged with ,

Game Dev Project Management Theorycrafting: Ownership Threads

May 5, 2021 Leave a comment

There are two dimensions to feature ownership in game development: how strongly are features owned by an individual member of each discipline, and how tight is the collaborative connection between those owners in different disciplines.

It is common to sort teams into pools of talent based on their discipline. This makes sense: two artists are much more likely to be able to share work than an artist and an engineer. From a project management perspective, knowing roughly how much “engineering” work and “art” work there is, measured against the number of “engineers” and “artists”, gives a good first approximation of how much time it will take to finish all the work for the game.

However, to a second approximation, interchanging tasks between members of a discipline carries a penalty, either in time or quality or both. One engineer may be more familiar with the game’s animation systems, while another is more familiar with the physics system. One artist may be faster at environmental modeling, while another is faster at prop modeling. This can be captured as “Ownership Strength”, the first dimension of feature ownership.

In a project with strong ownership, it is highly advantageous to silo your developers; that is, the engineer who works with the animation systems gets assigned all the animation-related tasks, and none of the physics-related tasks. In a project with weak ownership, each developer is able to take on any task within their discipline. Typically, strong ownership carries an efficiency bonus (devs get really good at working within their silo) at the risk of introducing bottlenecks (the game requires much more animation-related work than physics-related work, causing an uneven work distribution).

Each feature in a game requires some amount of work from each discipline. Artists, designers, sound designers, animators, and engineers all must collaborate to get a given part of the game to a shippable state. There are two broad strategies to managing this collaborative aspect. The first, as previously mentioned, is to sort devs into discipline-based pools, or “teams”. The second is to sort devs into feature-related “pods”, in which developers of all disciplines who work on a particular feature are treated as an organizational unit. This strategic spectrum can be described by the second dimension of feature ownership, “Collaborative Tightness”.

In a project with loose collaboration, an artist that asks for engineering support will create a ticket that is passed to the engineering team, which is subsequently assigned to an engineer, who will perform the indicated work and then send that ticket back to the art team. In a project with strong collaboration, an artist will work closely with an engineer, rapidly iterating and trading ideas, asking directly for support until the feature reaches the desired quality threshold.

Strong collaboration not only leads to higher quality features, but also stronger team morale due to a higher sense of teamwork and feature ownership. However, it comes at the cost of degrading the benefits of sorting developers into discipline-based teams. When small, agile, cross-discipline teams work iteratively and collaborate closely, it can be difficult to capture the work involved and effectively create a projection of how long it will take to finish all the work for the game.

That is, as long as we are using standard project management paradigms.

It is clear (to me, at least), that tending towards both Strong Ownership and Tight Collaboration is a recipe for a high-quality game. The only downsides are the risk of bottlenecking and the difficulty of predicting work schedules. These downsides are simply failures of the traditional game dev project management framework.

Starting with the first principles of supporting a Strong Ownership / Tight Collaboration development environment, let’s craft the ideal project management framework. This framework should ensure that the project meets internal and external milestones, is predictable in development time and cost, and meets the desired quality bar across all aspects of the game.

Strong Ownership is most effective when it grows naturally within a discipline. This accounts for natural collaboration between a group of artists, a group of engineers, etc. that leads to better problem solving and technical skill growth. Top-down assignment of feature ownership is a recipe for disaster. At the start of development, the Lead of each discipline should be considered the owner of all game aspects. As needed, the Lead can assign out ownership to specific team members, and similarly ownership can be passed between team members as needed.

As distinct aspects of the game become obvious and individuals begin gaining ownership over them, Tight Collaboration can be encouraged. In cases where it makes sense, an individual from each discipline can become the owner/liaison for that feature. For example, an artist has a single engineer, a single animator, and a single sound designer to talk to regarding a particular feature in the game. This group of inter-discipline collaborators can be called a “Thread”. Each person in the thread is responsible for getting the feature to the desired quality bar and estimating the required time to do so.

As development continues, more and more Threads will form, until there are very few aspects of the game that aren’t Strongly Owned by individuals. 

In many cases, there will be more work to do on a feature than can be done by a single person. Multiple members of a discipline can support the same Thread, either as primary owners or “secondaries” (in fact, having a secondary for each Thread would help reduce the “bus factor”). Tasks should be more-or-less assignable to any primary or secondary in a Thread; e.g., any art-related task can be assigned to any artist associated with the Thread. If too much specialization occurs within a Thread, it is time to split the Thread to form 2 or more new Threads. 

By polling each Thread, a project manager can quickly gain an understanding of how much work is required to bring each aspect of the game up to a given quality bar. For each person, the work from all of their Threads can be summed up and compared to find team members who are bottlenecks. The Lead of a discipline may re-assign ownership of Threads as needed in order to reduce bottlenecking to a minimum. High-level feature prioritization converts into Thread prioritization, which easily converts into task prioritization in whatever project management software is being used.

Within this framework, achieving a first approximation of the amount of “art” or “engineering” work is still completely possible. But a second approximation is much easier to get and much closer to the reality of feature development. We have unified team member management, schedule management, and feature quality management into a single theoretical framework.

Allowing feature designations to naturally precipitate out as Threads, rather than attempting to impose them top-down, also helps ensure the game reaches a quality bar across all features. For example, in a game where the player performs ranged and melee attacks on enemies, a project manager might think that “melee combat” and “ranged combat” are a good way to separate the different types of work. However, by allowing designations to precipitate naturally, the manager might discover the team finds it more useful to categorize work into “damage and dismemberment” and “weapon handling”. This sort of thing cannot be determined a priori, so allowing these feature designations to develop naturally via a formalized Thread framework is highly superior to imposing them top-down.

Overall, this sort of framework seems like a good compromise between a strict discipline-based team approach and a strict “feature cabal” approach, while maintaining more flexibility and accountability than either, and also boosting game quality and team morale.

Filed under Analysis, Blog, Games, Idea Tagged with , , ,

Will Half-Life: Alyx Set VR Back?

April 4, 2021 Leave a comment

Half-Life: Alyx poses a conundrum for the future of VR. On the one hand, it is praised as one of the best VR games yet. It is critically acclaimed and immensely popular, bringing a best-of-class experience to countless enthusiasts and VR-newbies alike. Yet, as some (disgruntled) people have pointed out, it hardly seems like the game is native to VR. Little in the game requires VR to be playable. The environmental exploration and the gunplay, arguably the core loop of the game, could be turned into a paint-by-numbers flat FPS.

How can we reconcile this paradox? If VR is truly a medium that can upheave and uplift interactive entertainment, how can its poster child be a simple shadow of a regular flat gaming experience? Will every VR game just be a watered-down pancake game with a small garnish of VR laid on top?

Previously, I posed two questions for determining if a game belongs in VR:

  1. Would I rather play this game in VR, or on a flat screen?
  2. If this game wasn’t in VR, would I choose to play it instead of another flat game?

The answer to the second question is almost certainly “No”. Flat games have had decades to refine their metaphors and mechanics; a game that has been shaped to fit in VR will inevitably lose out to non-VR games in their home territory.

So if the second question is answered “No”, the answer to the first question better be “Yes”. But it’s tricky to get a “Yes”. If you play VR games, you have likely felt frustration or boredom and wished that you weren’t wearing an uncomfortable headset and could instead snuggle on your couch and play on the TV using a gamepad.

But perhaps surprisingly, the answer is “Yes” for Half-Life: Alyx. Of course it’s subjective, but it seems that most people (myself included) are eager to strap on a headset in order to dive into that immersive world. What other games make you eager to throw on the headset? Only a handful. Beat Saber’s continuing popularity is in part due to the fact that user’s are willing to jump into the headset just to play a few rounds of Beat Saber.

There is no one answer to why Alyx succeeds where so many fail. Polished audiovisuals; excellent pacing in visuals and spatial design; finely-tuned tension in combat encounters; compelling characters, dialogue, and worldbuilding. The game keeps you present and immersed in the virtual world, and that is the value of the game. Conversely, the game’s core loop is sufficient, but not spectacular. You come for the virtual adventure, and the core loop keeps you from being bored.

Previously, I posited that a VR game can succeed only if it stands out on at least 1 of 3 pillars:

  • Kinesthetically pleasing core loop
  • Compelling character
  • Fantasy fulfillment

Alyx certainly succeeds at having compelling character. The worldbuilding, environments, characters, dialogue, and humor are all engaging. For fantasy fulfillment, the game leverages its distinguished IP, drawing you back into a universe that has been in our collective consciousness for nearly two decades, growing to near-mythic proportions.

And yes, its core loop has kinesthetically-pleasing elements. The tuning on the gravity pull is just right, making it a simple joy to fling objects towards your hand. The stress of reloading the pistol during a tense fight is exhilarating, when in other VR games it can be frustrating. Jabbing a syringe into yourself to heal while in cover makes you feel like an incredible badass.

It’s no wonder Alyx succeeded. It executes expertly on all the things that people seek in games, and VR games in particular.

And yet.

Mechanically speaking, the actions you do in the headset are things that can translate to non-VR. Point at a thing, pick it up. Reload a gun. Climb a ladder. Push joystick to walk. Point a gun, shoot. Crouch behind cover.

Compare to the acrobatic gun-fu of SUPERHOT, where you can dodge or block individual bullets, or the savage skull-bashing of GORN. Compare to the rhythmic, full-body synthesis of Beat Saber and Pistol Whip, or the graceful low-gravity navigation of Lone Echo.

Previously, I said that the promise of VR is one of hyperreality, a world that is better than reality: denser, more exciting, more enabling, less annoying. 

Half-Life: Alyx is hyperreal. But its delicate tuning is a brute strength solution to the problem. It isn’t through new clever design paradigms that Alyx captures the hearts and minds of VR gamers, but by relentlessly polishing a mediocre foundation. It is a triumph of budget, not a triumph of innovation.

And accordingly, history will not remember the game for its design. The other games mentioned above will accord a place in history as innovative pioneers, similar to how Spacewar, Pong, and Pacman are still fun to a modern gamer. But in 10 years, Alyx will hold little appeal for VR gamers. Game development tools will advance, decreasing the cost of fidelity. Its audiovisual polish will become less impressive. Correspondingly, its laggard design sensibilities will become more grating.

***

The sad truth is that there is an inverse relationship between design innovation and the overall scope of a game. When following well-worn paths, a development team can spend their energy creating more content and polishing the game. 

Gamers want big, polished games. VR is young enough that the bar has not yet been raised too high in terms of design expectations; players will suffer clunky design because, for the most part, they don’t realize it could be better. 

But clunky design limits the growth of the medium, because it restricts more casual players from being drawn in by the experiences. When an uninnovative yet polished game succeeds, it lures more developers into believing that it is fine to settle for less when pursuing the promise of VR. Collectively, the level of expectation is dragged down. The raising of the bar is stunted. And the longer the bar stays low, the longer new casual users are being exposed to an imperfect incarnation of hyperreality.

So Half-Life: Alyx makes me happy and sad. It is a triumph that gives visibility to the medium and may attract new users in the short run. But it is a mediocre vision of what VR could and should be like in the future. It is sufficiently hyperreal in today’s marketplace, but it accomplishes this with brute strength, not with cleverness. And in doing so, it threatens to set the medium back in the long term.

Filed under Analysis, Blog, Games, VR Tagged with , , , ,

Real-time Strategy in VR

February 22, 2020 Leave a comment

Real-time strategy is the genre that’s possibly been knocked hardest by the evolution of modern gaming. Interest in games like Starcraft have waned in favor of action games and MOBAs. There are two competing explanations for this: either the modern market isn’t big enough to support an RTS ecosystem, or the developers of real-time strategy games have failed to innovate and keep up with the times. In either case, it is undeniable that the RTS is a dead genre.

Or is it? Of course not! The spirit of the RTS is alive and well. The same things that hooked players back in the 90s still hook players today. Human psychology hasn’t changed. But to survive in the vast and confounding battlefield of modern gaming, the genre has had to twist, split, and adapt to fill sustainable niches. To the legions of fans forged in the heyday of the RTS, it seems as though the genre is dust and bones because the RTS is held, in the public mind, as a single monolithic conception.

The RTS suffers the same fate as Star Wars. A novel presentation in a fallow market captured the hearts and minds of many, for myriad reasons. Unfortunately, its fame necessitates its failure; the intersection of so many interests leaves nowhere to progress, creatively. Moving in any particular direction will cause some fans to lose interest. So in a sense, the RTS genre is dead, if you define the RTS genre by the specific mechanics found in Starcraft, Warcraft, and Dune. Such a precisely defined genre is a dead-man-walking.

To find the modern RTS, one must look at the aspects of play that create such a devoted fanbase. As I mentioned, these are diverse — from a quick survey of internet threads where people discuss why they love real-time strategy:

  • Building power over time
  • Introduction of mechanics over time causing increasing complexity
  • Discovery / Exploration
  • Having to choose where to invest time and resources
  • Managing multiple tasks / dividing attention
  • Single-player stories
  • Challenging yourself
  • Directing troops
  • Base building
  • Devising strategies offline and then implementing them ingame
  • Improvising when a plan falls apart
  • Defeating an equally-matched human opponent
  • Analysing your own mistakes and better learning the specifics of the game to improve

There are some high-level generalizations about player experience to draw from these. I think the attractions of a traditional RTS can be reduced to the following:

  • Fantasy of Command / Warfare
  • Struggling to acquire and subsequently manage streams of information (in real time)
  • Efficiently growing a base, set of units, and pool of resources (in real time)
  • Overcoming opponents by building and manipulating a superior mental model of the game dynamics (in real time) [“dynamics” in the sense of the MDA framework]

Note that I have appended “(in real time)” because these can be fulfilled by a number of turn-based strategy and tactics games, just not with a time-constrained component. Indeed, what separates the RTS from other “games of command” seems to come down to giving commands at multiple locations at the right time. Many modern games of command involve a single unit, or do not have a relentless time component (i.e. not turn-based and no pause feature). For this reason, the “Fantasy of Command” can be safely put aside for the purposes of this essay. It is fulfilled by other games and is incidental, not essential, to the genre.

We expect that modern inheritors of the RTS mantle will continue to fulfill some subset of these general player experiences. Indeed, we see Offworld Trading Company fulfills most of these, but is weak when it comes to “Fantasy of Warfare” and “Struggle to acquire information”, since the game lacks any units, and there are few mechanics that allow players to learn more about their opponent than what their opponent knows about them. Clash Royale delivers a strong experience when it comes to “Overcoming opponents through superior mental models”, but lacks virtually any information-gathering or economic growth components.

Those are the elements of the high-level player experience, the aesthetics of play. When it comes to crucial mechanics, I found this analysis compelling. The analysis essentially pinpoints 11 features that are both necessary and sufficient for calling a game an RTS:

  • Players themselves are not in control of turn progression, and there are no direct interruptions of the progression of actions within the game.
  • Not being in direct control of the pacing of game events put pressure on the player to make fast, accurate decisions based on limited information.
  • Poor decisions must be eliminated or mitigated in future attempts (that is, later in the course of a particular match or in future matches against the same or different opponents).
  • RTS train players to quickly evaluate situations to determine the best future path forward.
  • Acquisition and expenditure of stores of value
  • Array of options with which to progress
  • The player must be asked to invest limited (though not necessarily scarce) resources into progressing and expanding, capitalizing on their past actions towards future goals
  • Players be able to actually lose their investments
  • they must defend their own investments and use them as wisely as they are able.
  • Require the player to simultaneously manage multiple game pieces or elements.
  • Uncertainty of other player’s actions to be incredibly important

As a summary:

Multiple participants engage in competitive economics, managing limited resources to expand multiple game elements in order to gain an advantage and ultimately wrest control of one or more critical systems to attain a concrete victory.

While this is useful for determining what is and isn’t an RTS, or figuring out which elements of a specific RTS are core to its being and which are secondary embellishment, it doesn’t necessarily lend itself to the sort of abstract, blue-sky thinking that you need when pushing an existing, well-loved concept into uncharted waters. You could follow its prescriptions to the letter, and find yourself with a product that fails to fit its niche or entertain its users.

Real Time Strategy in VR

Plenty of developers have tried to make archetypal ports of existing genres into VR. These evidence themselves as being bids for the statement “X is the Y of VR”. “Pavlov is the Counter-Strike of VR”, “Space Junkies is the Quake of VR”, “Sprint Vector is the racing game of VR”, “Beat Saber is the rhythm game of VR”.

A good rule of thumb here is to ask, “if I had the choice of playing this in VR or not in VR, would I rather play it outside of VR?” Then ask, “if I played this as a non-VR game, would I rather be playing something else?” The answer to the second question is almost always “yes”, so the answer to the first question better be “no”.

Everybody wants a VR strategy game, and plenty have been made: Tactera, Base Blitz, Airmech Command, Brass Tactics, Skyworld, Landfall, Cosmic Trip, Final Assault. (I’ve played all of them, by the way). Plenty of these come close to being an archetypal port, a true “RTS of VR.” Personally, I’m currently enjoying Final Assault.

But would I rather play these outside of VR? Absolutely. VR is uncomfortable, and these games don’t (for the most part) bring anything to the table that precludes playing them on a flat screen. No game has yet provided an RTS experience that is integral to VR.

Can we define an envelope for the “ideal” VR RTS? I will suggest two heuristics that help us towards a vision of this hypothetical game.

Heuristic 1

The perhaps less controversial heuristic is that the game shouldn’t be worse off for being in VR. By this, I mean that the game is not less player-friendly or less fun than the “pancake” version of the game you would get if you tried to port it back to traditional platforms. This applies to qualitative things like “fun”, but also concrete things like input level and game feedback. If the input scheme of the game is frictional, the player will immediately reject the game. The controls should grant the player new opportunities, not restrict their ability to act.

There is a corollary which follows logically from that first heuristic: the inputs to the game should not be able to easily be mapped to mouse and keyboard. For example, all of the aforementioned VR games — with the exception of Cosmic Trip — involve commanding troops to move around on a 2-axis battlefield. There may be height variation, but there aren’t even multiple levels. In addition, troops are commanded by selecting groups of units, then directing them to a point on the battlefield. Naturally, this is exactly how pancake RTS games work, and it becomes quickly obvious that a mouse and keyboard is much better than a pair of VR controllers for this kind of work.

But this also applies at a higher level of abstraction. It may not seem problematic if my VR RTS involves giving commands to troops at ground level by making gestures. How could that be translated easily to a mouse and keyboard setup? Well, players tend to build the most abstract mental model possible when learning a game. Unless there is a significant amount of additional unique control afforded by this gesture-based command system, there will be no difference between it and a simple top-down point-and-click command system in the player’s mental model of the game. It would be difficult to build a gesture system that affords control so unique that it couldn’t be easily replaced by a pancake GUI using a few buttons, hotkeys, sliders, or mouse gestures. But trying to imagine such a nuanced gesture-command system might lead to some interesting RTS ideas!

Heuristic 2

The first-person, gesture-based RTS as a thought experiment points to an important distinction to be made between the hypothetical “ideal RTS” and existing archetypal ports of other genres. All successful archetypal ports are currently action games. At a most basic level, the game’s fun is grounded in a certain viscerality. Arizona Sunshine isn’t about learning the game’s sandbox; it’s about being in a horror situation, about reacting to surprises and danger under loads of stress and anxiety. Sprint Vector is a game of going fast and perfecting your execution of maneuvers. You beat out opponents by performing actions more precisely, not because you’ve built a more sophisticated mental model.

Conversely, three of our four identified core player experiences of the RTS genre are dependent on the fact that the player will be trying to understand the game as thoroughly as possible (i.e. build a mental model):

  • Struggling to acquire and subsequently manage streams of information (in real time)
  • Efficiently growing a base, set of units, and pool of resources (in real time)
  • Overcoming opponents by building and manipulating a superior mental model of the game dynamics (in real time)

A good mental model will help you filter, prioritize, and categorize the information you collect. It will help you manage your resources most effectively, and it will allow you to overcome your opponents. The core fun of an RTS is the feeling of having achieved victory by being a total genius. The core fun doesn’t come from ordering troops around on a battlefield.

This leads to a second heuristic: victory in the game should be determined by whoever builds a better mental model. It should not be determined by whoever masters the controls better, or acts faster. This heuristic also ensures that one of the most egregious VR issues is avoided: frustration at the controls. It can be stated with certainty that if the player is annoyed at the controls of a game, the designer has failed. Having the player fail because they are struggling with the controls is the worst possible outcome.

Constraints For An “Ideal” VR RTS

We can take it as a given that this hypothetical game is about achieving victory. Thus, it must contain mechanics that combine to produce interesting, nuanced dynamics, since the core gameplay is the way the player uses their mental model to understand the dynamics, and subsequently determine the best way to harness the mechanics to their end. Additionally:

  1. The controls of the game cannot be something which map easily to flat gaming (mouse and keyboard, touchscreen, etc), following from Heuristic 1.
  2. It is difficult to imagine a mechanic which would work well in a pancake game, yet whose corresponding controls do not map well to mouse and keyboard (or other flat input device).
  3. Therefore, any mechanic essential to the game must be something which could not easily exist in a pancake game.

This is a tall order. It calls for something few VR games have achieved, which is a core loop and core mechanics that cannot exist outside of VR. We need, essentially, the Beat Saber of the RTS genre.

When we envision “VR native” mechanics, we must start at first principles. What does VR have at its disposal? The ability to look in all directions; locate audio sources; ability to feel presence in a space; a sense of scale; room-scale movement (the ability to crouch, jump, and walk around a space); and tracked hand controllers with haptics.

Acquiring Information

We could use the perceptive aspects of VR (looking around, binaural sound spatialization, stereo vision, haptics) to feed into the “struggle to acquire and manage information” player experience.
But we need to be careful not to violate Heuristic 2. For example, in Brass Tactics, the fact that you can’t be everywhere at once is an important gameplay element. You might think this would force players to use their mental model to choose the best place to focus their attention — a positive gameplay element. However, in reality, it means that the person who is better at moving across the table and orienting themselves using the game’s controls gains the advantage. The player who is better at using the controls is more effective.
In our ideal RTS, players must be able to quickly take in all the available information using an incredibly intuitive interface.

We should keep in mind the current limits of headsets. Resolution is very low (compared to angular size of pixels on flat displays), and focal length is fixed to a single distance. Empirically, visual acuity is reduced to between 20/32 and 20/42, depending on the anti-aliasing and supersampling settings the user employs [1]. Oculus says in its best practices document that it is most comfortable to view objects between 0.75 and 3.5 meters away, since the focal distance of the headset is about 2 meters.

Here is an image demonstrating a vergence mismatch between eyes.

Accordingly, we should place all important elements of the interface 2 meters away from the player (this minimizes the vergence-accommodation mismatch problem, illustrated in the image above). Any text (which there should be very little of, since people hate reading) needs to be big enough that someone with 20/42 vision could read it comfortably at a distance of 2 meters.

Managing Resources

As the player deals with resource management (let us generalize structures, units, and resources pools under the term “resources”), they will contend with VR’s input methods. At a low level, this is the position and rotation of their head and hands, and the buttons on their controller. The pose of the head should not be used as to not interfere with the player’s information gathering, so we are left with the controllers. Here, there are too many possibilities to enumerate.

Most of the existing VR RTS games utilize variations of the “laser pointer” input paradigm, which ultimately boils 6 degrees of freedom (DoF) into 2 dimensions (specifying a point on a planar topology). Our ideal RTS would require input along at least 3 dimensions, increasing the difficulty of mapping it to a accessible pancake and thereby increasing the likelihood of satisfying Heuristic 1.

Our input interface must also contend with Fitt’s Law. Specifically, in VR players can either do a lot of sweeping actions quickly, or they can do few precise actions slowly. If the interface requires precise actions, and increasing action frequency increases player power, then a player will become frustrated as they try to actions faster than is possible for them. However, even if the inputs involve broad movements, we would still like to avoid coupling input frequency and player power.

(As an aside, games like Starcraft contain just such a coupling. The faster and more precise your actions, the better you will do against a slower opponent of equal strategic mastery. However, this is archaic. Other modern genres have captured this “twitch” aspect. The modern RTS may safely abandon it.)

If inputs fall on the slow-yet-precise side of Fitt’s Law, they should not require fine wrist control. Many VR games use a laser pointer-style interaction, and anybody who has played these games knows that hovering over a small UI element and pressing a controller button is an exercise in futility. Besides frustrating the player, encouraging myopic focus on small interface elements means the player won’t be focusing on the larger environment around them, an experience that is one of the key selling points of VR.

Putting It Together

To recap:

  • We should avoid coupling input frequency and player power.
  • Inputs should not require fine wrist control or angular precision.
  • Inputs to the game should have 3 or more dimensions.
  • Interface elements should sit around 2 meters from the player, and not require much visual acuity to operate.
  • Players must be able to quickly absorb available information using an intuitive interface.
  • The mechanics must combine to produce interesting, nuanced dynamics.

Additional easy wins would include a focus on co-presence with other players, a component of play that involves scale (seeing big things or feeling big in VR is cool), and the option of playing the game sitting.

Side Note: Ergonomics

Many VR RTS games have poor ergonomics. Since their action takes place on a flat horizontal plane, you spend the majority of your time looking down. With the weight of the headset on the front of your head, this induces considerable neck strain over time.

Additionally, a lot of the games use some form of world-dragging to translate the player around the virtual game space. Smooth artificial locomotion causes people to stand still, which is biomechanically more tiring than moving around on your feet.

Our ideal game would have you looking straight or slightly up during the majority of the game, and would take place within a fixed volume to encourage the player to move around physically.

Specific Ideation

This section is going to be a bit of a freeform brainstorm on how to drive the design of our hypothetical game.

The geometry should be topologically 3D and exist within the player’s real space volume. Perhaps the battlefield exists as a series of nodes with links between them. Nodes would provide both resources and 3D arenas for engagements, while links could be created and destroyed over time, and have different properties (slow/fast, dangerous, etc).

In order to decouple input frequency from player power, we can implement an action budget. This might be a power meter that slowly fills up, and allows manipulation of nodes and other resources on the board. Managing and budgeting your energy would be an important meta-game. You could burn energy to push an offensive, or burn energy to defend more effectively. Having low energy would make you vulnerable. Final Assault is an example of a game that does this well.

During direct engagements, when two players are both competing for the same location on the battlefield, we must make sure to provide significant choices to players and also prevent the faster player from automatically gaining the advantage.

The majority of an attacker’s effort would be in preparing an attack. Adding a delay between finishing preparations and the actual attack would allow defenders to be alerted and turn their attention to the engagement.

Micromanagement would be penalized – it could cost energy from the action budget, and take time to be delivered to the battlefield. Sending too many commands in succession could paralyze your units.

Leveraging high-level inputs like unit formations and group composition could be effective. Having unit behavior change significantly based on group composition could allow deep strategic play and forward-thinking. For example, assigning a fighter escort to your bomber squadron could cause the bombers to focus on making headway towards their target rather then spend effort defending themselves.

This concept of purchase-then-attack could be expanded by allowing players to purchase reserve units at a discount; if you can properly judge how an engagement will play out, you can get proportionally more power onto the battlefield in the right place at the right time.

Resource income should be expandable, but not exponentially. We want to prevent “steamrolling”, and instead foster a tug-of-war dynamic. Spending time arranging your resource production in 3D space could yield some efficiencies — but ideally this requires a trained eye rather than a fast hand.

In addition, adding unit exhaustion could prevent a well-designed push from completely overrunning the opponent without giving them opportunity to counter. Making defense easier than offense means you need to continually exercise superior play in order to win. Over time, this balance should become more unstable – this prevents a complete stalemate. After many minutes of play, a single advantageous play could cascade to victory.

Perceivable situational details should influence the outcome of encounters. Things like group composition, terrain, weather, flanking, formations, and unit morale, if properly exploited, can lead to one-sided engagements.

Game conditions should change to prevent a player from establishing a totally unassailable position. For example, a changing battlefield topology would open new flanking routes or render previously vital positions redundant. This would also require players to continually adjust their personal strategy, leading to more interesting matches.

We want to encourage strategic posturing. Placement of units should be a mind-game, to some extent. Direct engagements should resolve fairly simply and quickly, discouraging micro-management and feelings of helplessness as your forces lose. There could be several discrete points during an engagement where player commands are communicated to the units – creating a rock-paper-scissors guessing game. “Is the opponent going to change his formation command, or keep it the same?”

All together, this should lead to a “dance”, where the player shifts around the space, setting commands up that will execute at some time in the near future.

Anyways, I hope this has excited some ideas in your brain. It has certainly done so for me.

Filed under Analysis, Blog, Games, VR Tagged with , , , , , , ,

VR Games are Less Immersive than Pancake Games

February 1, 2020 1 Comment

When people first try VR, they often experience The Blu. It is a spectacular demonstration of the presence and immersion possible in VR. In the most popular sequence in the experience, you find yourself on the deck of a sunken ship. As you marvel at the beauty of this underwater environment you have been transported into, a gigantic whale comes into view, mere meters away from you. It pauses to eye you (curiously? balefully?), before swimming away and sending a rush of water past you.

Wow! The fidelity of the environment in its visuals and audio stun the senses at first. Then, you realize you can walk about this deck as if you were there, even crouching down to inspect objects and fish, or reaching out your hand to brush the fauna of this seascape. Now, you come face to face with a creature whose scale you have only been privy to before maybe once, in a natural history museum. But this whale is living; you can lock eyes with it, and the encounter is as ephemeral as a real encounter with a wild creature – before you know it, it is gone.

This is VR in its best form: you are truly transported to a realm that is better than real life. Even if you took the time to become certified for scuba diving and started exploring sea wrecks and swimming with sea life, it wouldn’t be the same. VR allows you to strip away the scuba mask and the hours of training, the cost of taking a boat out to a site, and the danger of entering an alien world. It condenses a transcendent experience into a package that is available to ANYONE, even the young and the handicapped. It is hyperreal: better than reality.

But most of what you can experience in VR does not match this level of hyperreality. It struggles to justify the friction of the medium – the setup time, the cost, the discomfort of strapping a device to your face. Even enthusiasts soon realize that the available virtual worlds of VR don’t offer a better experience, holistically, than sitting on the couch watching Netflix or playing a 3rd person action game on the computer. Why is this? There is a fundamental calculation being performed unconsciously:

Relative Value = Unique Benefits – Unique Downsides

VR has a lot of unique benefits as compared to traditional pancake gaming, as well as some unique downsides. One large benefit is the novelty factor. But this benefit decreases with exposure, and in the end this calculation of relative value results in a negative number for most people. The data backs this up. 40% of people only use their VR device once a week, 34% of people use it less frequently than that. [1] A lot of people mostly break their device out to show to someone else.

Some of the common unique benefits of VR:

  1. Presence – VR has a phenomenal ability to make your brain believe you are truly in a different space.
  2. Massive input space — 6 DOF VR’s motion controllers afford many degrees of freedom over traditional computer inputs. Not only are there the eponymous 6 degrees of freedom (position and rotation), but there is linear and angular velocity, two or more analog inputs (joystick, trigger, etc), and several digital inputs (face buttons).
  3. Physical Freedom – you can transcend the limitations of your mortal form, flying across the world like a superhero or diving deep into the depths of an abyss.
  4. No Consequences – this is a benefit shared, nominally, by traditional games. In Grand Theft Auto you can shoot, steal, and drive like a demon without consequences – you won’t feel bad for killing real humans, and you won’t go to jail for the rest of your life. In VR, you can explore the same kind of consequence-free space, but mapped much closer to reality. You aren’t pressing a button to smash open your opponent’s skull in the gladiatorial arena; you are actually doing it!

Some typical downsides of VR:

  1. Time, Space, and Monetary Cost – not only are the headsets and computers expensive, but you often need to dedicate space in your house to them, and spend time setting it up and maintaining it.
  2. Hassle – by this, I mean the stress imposed by the ensemble of equipment. In order to enjoy the benefits of VR, you must contend with adjusting the headset to fit your head, finding a proper IPD setting, adjusting the headphones or earbuds, picking up the controllers after putting on the headset, adjusting straps on the controllers, minding the headset cable during play, avoiding lens fog, etc. Then sometimes the computer will be acting up, requiring some troubleshooting and application restarting.
  3. Discomfort – even with a perfectly situated headset, it presses against your face and scalp and heats up. Most devices also introduce some ocular discomfort over time, whether due to pupil swim, IPD mismatch, or other subtle optical problems.
  4. The Experience Mapping Problem.

The Experience Mapping Problem

The human brain is fantastic at recognizing patterns and drawing connections. As young children, we gain an intuitive understanding of physics by observing how objects react to our inputs – we unconsciously construct an elaborate mental model of reality, which allows us to accurately predict the outcome of our actions. This is what lets you grab a mug and set it down elsewhere without spilling its contents, or push through a door and enter the room beyond. Sometimes, this model is incorrect: you misjudge the weight of the mug, or the door is locked. When this happens, your body often continues to execute a planned sequence, causing you to drop the mug or run into the door. Trained on hundreds of thousands of hours of experience, our mental model maps certain stimuli to certain responses – and it takes a lot to break that linkage and reform your model to account for a new reality:


VR, fundamentally, runs smack into this mapping problem. Traditional pancake games exist as a separate reality – one that exists on a screen, and which you interact with by pressing buttons. You construct a new mental model when playing pancake games, one that maps the stimuli on-screen to button-pressing responses. Virtual reality, on the other hand, intends to present a reality that mirrors real life. You see it and hear it the same way you see the real world, you can move around with your actual body and you can use your hands to interact with the world – just like real life!
So your brain, naturally, attempts to use the same mental model to react to stimuli in VR. But this rarely serves you well. Current digital worlds function with radically different rules. And this radical mismatch between your mental model’s expected outcome and the observed outcome breaks immersion.

This is the basis behind my claim that VR is actually less immersive than traditional interactive simulations. This runs contrary to what you would expect; what could be more immersive than actually being physically present in another world? But what is immersion? Immersion is the transportation of the spirit – when your subjective experience is completely subsumed by a piece of media. You can be sitting on your couch with a controller, but be completely immersed in the fantasy action game in front of you. Your entire consciousness is in the world described by the image on-screen. Yes, your bodily presence never leaves the couch, but your mind and soul are elsewhere. This can happen even with a movie or book. A book is not naturally immersive, but eons of narrative craft can be leveraged by a good author in order to transport you to another time and place.

The antithesis to immersion is any reminder of your bodily presence in the real world. If you must consciously acknowledge the duality of your existence (the body in one place, the mind in the other), your mind is drawn back to the mortal coil. A distraction causes you to look up from your book, someone walks in front of the TV, or a phone goes off in a theater.

On one hand, VR enables a powerful sort of pseudo-presence. Your eyes and ears are physically present in this other reality, as are your hands (sort of). On the other hand, this physical presence is hampered by invasions of external elements — the cables, the screen-door effect, limited field-of-view, inner-ear and proprioceptive discrepancies — which remind you that you are strapped into a headset.

In addition, many VR experiences are chock full of metaphorical phones-in-theaters. The unconscious attempt to apply your mental model of the real world to this virtual world results in constant discrepancies between expected and observed results, which must be resolved with conscious effort. This further drags your mind away from the constructed reality and back into real life.

What can we do? How do we create a virtual reality that delivers on the promise in The Blu?

How do we tip the calculation of relative value from red to black?

Simply posing the question suggests an answer: we leverage the unique benefits and minimize the unique downsides. Fortunately, lots of people are already working to this end. The monetary cost of VR is decreasing, along with the setup and hassle. Technological improvements in display technologies and audio simulations boost presence, and new apps give players new freedoms and consequence-free fantasy fulfillments.

People are working on these problems because they are obvious. But the Experience Mapping Problem is not obvious. Its results are obvious – a lack of immersion. But this is frequently diagnosed as a fidelity problem, to be solved by higher resolution displays, more realistic graphics, advanced physics simulations, and more “natural” controllers. Unfortunately, these things may actually worsen the Mapping Problem with current-gen VR.

[Asgard’s Wrath]

[Job Simulator]

Which is more immersive? Asgard’s Wrath is more realistic, but you are less likely to lose yourself in the simulation.

[Hand tracking]

[Oculus Touch]

Which is more immersive? Finger tracking is more realistic, but when you try to grab a virtual object, the outcome is VERY different from the expectation based on a lifetime of using your hands.

One solution is to create a set of stimuli that are so different from reality that the player does not make the mistake of assuming their IRL mental model will apply in this space. But this means we must abandon the benefits of VR related to player fantasy. Physical freedom and consequence-free spaces are less meaningful if they are completely unrelated to reality anyways. Nonetheless, some experiences like Tilt Brush and Oculus Medium leverage the input freedom and physical movement freedom to create engaging abstract experiences. But the promise of a hyperreal VR world that is “like real life but better” demands that we try harder to resolve the Mapping Problem.

There are two paths to resolution – the first (and more common approach) is to expedite the player’s development of a new mental model. The second path is to develop virtual realities that are designed to gracefully accept players’ actions and minimize mismatches between expected and observed outcome.

We need elements of both solutions in order to achieve VR’s potential — but modern games need more of the second path. Forcing the player to develop a complicated new mental model to interact with virtual reality means sacrificing a significant amount of VR’s hyperreality, and thereby decreasing its relative value. But we need to maximize the relative value of VR — it’s now or never. If VR doesn’t enter the mainstream now, it will likely never flourish.

The most popular apps and games can teach us how to minimize the mapping problem; games like SUPERHOT, Job Simulator, Robo Recall, and Thrill of the Fight.

Let’s make VR the immersive hyperreality we were promised by science fiction:


(video source)

[1] “77% of Virtual Reality Users Want More Social Engagement”

Filed under Analysis, Blog, Games, VR Tagged with , ,

What can the current VR market teach us about design?

October 4, 2019 Leave a comment

In this post:

 

Introduction

I figured I would write this post now, since it is rapidly becoming outdated. For a while now, I’ve been following the popularity of various virtual reality (VR) games. Specifically, I’m interested in the real player engagement generated by these games, for the purpose of creating a rough qualitative model which can predict a given game’s success.

What are the stakes here? Assume we want to make a profitable VR game on a (relatively lean) budget of $500,000. In order to break even, you need to sell 25,000 copies at a $30 price point. In reality, your price point is probably lower ($20 or $25), and you’ll be selling a large number of copies at a discounted price point in Steam sales. Your budget may also be higher (think 1-2 million dollars). However, we can also assume your sales figures will be roughly doubled if you release the game on PSVR, and maybe tripled if you release on Oculus Quest (big assumptions, but these are all ballpark numbers anyways).

More than 1,000 games with VR support were released on Steam in 2018. Even assuming that 80% of those are hot garbage, you need to beat about 165 other games in order to reach that 25,000 sale mark. That’s right: by the time you get to the 35th best-selling game in 2018, you are looking at games that only sold around 25,000 copies.

I will delve a little more into the specifics later, but the crux is that, in 2018, there is heavy correlation between sales and active playerbase. That is, the games that people keep playing tend to get the most sales. Therefore the pertinent question is: what can we glean from the top-played games, so that we can more reliably develop profitable VR games?

More explanation can be found at the end of this post. But, a caveat: here I’m mainly looking at Steam and assuming it is a representative sample of the market at large. Ok, let’s jump right into it.

 

What are consumers choosing?

If we take a gander at what games people are playing on Steam in a given day, you’ll find a list that looks like this:

  • Beat Saber
  • Pavlov VR
  • B&S (full title: Blade and Sorcery)
  • Rec Room
  • H3VR (full title: Hot Dogs, Horseshoes, and Hand Grenades)
  • Skyrim
  • Arizona Sunshine
  • Job Simulator
  • SUPERHOT VR
  • Onward
  • Fallout
  • GORN
  • Elven Assassin
  • The Lab
  • Zero Caliber VR
  • Space Pirate Trainer
  • STAND OUT

This is roughly ordered by player count. Beat Saber usually has between 1000 and 1500 players, Pavlov usually has about half that; Rec Room, B&S, and H3VR have a few hundred players, and the others have between 120 and 20 players.

(Rec Room and The Lab are free, so we will ignore those henceforth)

Most of these have between 200,000 and 500,000 owners on Steam. Some are lower; B&S, Elven Assassin, STAND OUT, and Zero Caliber have 40,000 – 100,000 owners. Note that you can’t multiply the owner count by the sale price to get the gross revenue, because owners include people who got it for free or at a heavy discount. However, high owner count usually means high revenue.

All of these “top-played” games landed on the top 20 best-selling list of games for 2018. Despite the low ratio of active players to total owners, the top-played list is remarkably stable. All this suggests that it isn’t freak chance that these games are on top.

There are only a few games from the top 20 best-sellers in 2018 that aren’t on this “active playerbase” list:

  • Orbus VR
  • DOOM VFR
  • Raw Data
  • Rick and Morty
  • I Expect You To Die
  • Budget Cuts
  • Sairento VR
  • Sprint Vector

These all have ownership numbers between 20,000 and 100,000 on steam, but normally 10 or fewer active players at a given time.

The ones in bold are linear singleplayer games — i.e. you play them once and you’ve gotten everything out of them — so it isn’t surprising they don’t have an active playerbase. Three of the others, OrbusVR, Raw Data, and Sairento VR, benefit from a first-movers advantage. OrbusVR is the first “VR MMO”, while Raw Data and Sairento were some of the first games with significant amounts of content and “good graphics”. This has placed them as well-known titles in the VR market, and continues to drive consumer interest despite the fact that they clearly can’t sustain player interest. I would argue that Sprint Vector also benefits from a sort of second-hand first-movers advantage, being developed by Survios, the same company behind Raw Data (and thus benefitting from higher consumer awareness).

Many of the top-played games also benefit from a first-movers advantage. Some happened to be high-quality games in a very early market: Space Pirate Trainer, Arizona Sunshine, Job Simulator, SUPERHOT. Some happened to hit a particular niche, maybe without being high-quality: STAND OUT, H3VR.

But, are there features intrinsic to these games that we can learn lessons from?

Only 40% of the top-played games have a multiplayer mode, and you would expect games with active playerbases to have multiplayer support. When you consider all the top 35 best-selling games (which, remember, you need to be in to turn a profit) only 25% have multiplayer. I used to think that a VR game needed multiplayer support, even if people didn’t tend to use it, because it added significant perceived value to the consumer. This is clearly not the case. From a numbers perspective, adding multiplayer support is currently not worth it.

I suspect that having a game with an active playerbase is healthy for sales, since it places you on the front page of the “What’s Being Experienced” chart in the Steam store. It looks like this:

As a player, this list is very appealing. It shows what games people have found to be continually fun; if I buy a game from this list, I have a higher chance of maximizing the bang for my buck.

So while VR games can and have been successful with low replayability (Moss and I Expect You To Die come to mind), creating a game that players can return to night after night significantly increases the chances of making a profit. A game that people can keep playing is also a game that people will keep talking about both online and in real life, and word-of-mouth is not to be underestimated as a force for generating sales in the VR market.

However, it is incredibly difficult to provide the sort of value that keeps a player entertained for months, especially since every player wants something different. This is where user-created content and mods are invaluable. It is no accident that the games with the biggest active playerbases are also well-known for their user-generated content and mod support: VR chat, Rec Room, Beat Saber, Pavlov VR, B&S, and Skyrim. In fact, it could be said the mods for these games are more popular than the games themselves.

If someone says “I couldn’t imagine playing this game without mods,” as is often said of those games mentioned above, it isn’t a sign of failure on the developer’s part. It is a sign that they have provided a platform that will continue to excite people and generate sales, even without further effort from the developers. It is the holy grail of VR game development: maximum engagement at minimum cost.

Thus, we have a basic template for thinking about a VR game with a chance of profitability: a mod-friendly singleplayer game that a player can jump into night after night and that makes them want to talk about their experiences.

 
 

Three Design Pillars

What keeps players coming back? What keeps them in the headset when they could be doing other things? In my estimation, all the top-played games succeed in one or more of three design categories, or pillars:

  • Kinesthetically satisfying core loop
  • Colorful and compelling atmosphere or character (henceforth “compelling character”)
  • Fantasy fulfillment

 

A kinesthetically satisfying core loop is a basic gameplay loop that, absent all else, makes you move your body in a way that feels good. The best of these have the player doing things you can imagine a kid doing by himself on a playground just because it’s fun to do. Beat Saber, B&S, GORN, Space Pirate Trainer, and SUPERHOT all get you moving in satisfying ways. There isn’t a lot of standing still, trying to point your controller at something, or fumbling with menus, or fiddling with two small objects. They have sweeping motions and encourage you to sway your body smoothly and sinuously. When describing the game to your friends, you can move your body and make sounds with your mouth to convey the experience.

Compelling character is when a character in the game, or simply the attitude of the game itself, makes you want to stay in it. Arizona Sunshine has a fun self-narrator that lends life to a game that otherwise would become a drag after half an hour. Job Simulator is silly and absurd. GORN is a masterful blending of comical and gory action that sucks players right into the universe with minimal friction.

Finally, fantasy fulfillment is the thing most players actively look for in a VR game. They want to be a Jedi, a gladiator, a marine, a wizard, a survivor of the zombie apocalypse. Whether through the story, the action, or the environment, a game with fantasy fulfillment transports the player to a different time, place, and role. Their return to the real world after a play session is a shock, and it creates a yearning to return to that place where they were something different than they are in real life.

Some VR games ride solely on their fantasy fulfillment. People harp on Skyrim VR for being a bad VR port, but it hits the top-played list because it has such rich, immersive environments. H3VR is a gun simulator with some game modes tacked on. Most successful VR games have at least partial elements of fantasy fulfillment. Even Beat Saber, a game that isn’t really *about* anything, still generates fawning comments about how it really feels like you are wielding a lightsaber.

Obviously, each of these pillars are highly personal. Different people like different characters, have different fantasies, and enjoy different motions. For example, I can’t stand the bow-and-arrow motion in VR, but I know a lot of people enjoy it — hell, the only VR “genre” more prevalent than bow-and-arrow games are shooters.

It is thus a developer’s goal to execute on a concept that squarely hits all three pillars for their target audience while still doing the other things a game needs to do to succeed, like providing a unique value proposition to the player and being easy to market. These three design pillars are necessary, but not sufficient, for success.

Does this describe your VR game?

A single-player game that fulfills a fantasy for players. Once in the headset, it immediately captures players with a kinesthetically fun core loop, and keeps them playing for its compelling character. Players want to talk about their experience in the game and play it again, exploring user-generated content and mods to play exactly what they want and how they want.

 
 

(The rest of this post is data sources and housekeeping. Feel free to skip it.)
 

Appendix

I’ve been using data from a few sources:

https://vrlfg.net/ VR LFG provides live stats from Steam for VR games.
http://steamspy.com/ SteamSpy provides historical data and ownership numbers for Steam games.
https://vrscout.com/news/steam-leak-reveals-vr-player-count/ This was a player count leak in summer of 2018.
https://store.steampowered.com/sale/2018_top_vr/ This is a list of games by “top-selling in 2018”, measured by gross revenue, sorted into buckets or “tiers” (but not ordered within a given tier).
http://gamstat.com/games/ Not relevant to this post, since I focus only on Steam stats, but GamStat provides stats on Playstation games including PSVR, currently the largest virtual reality platform.

The numbers I used in this post are mostly from May 2019, but I don’t think moving that needle backwards or forwards by 6 months would change the conclusions of this post.

To put game owner counts in perspective, at the tail end of 2018 there were roughly 2.5 million Oculus Rift headsets sold, and 1.5 million Vive headsets. There were also around 4 million PSVR headsets (thus the comment about releasing for PSVR doubling sales numbers). These PCVR numbers from Statista are corroborated by a report by NVIDIA that there are about 4 million PC headsets total out there.

[1] PCVR headset sales from Statista
[2] NVIDIA PCVR headset count confirmation
[3] PSVR headset sales

There are two major storefronts on PC — Oculus and Steam. This hampers analysis a little, because numbers from Oculus are basically impossible to come by. However, based on some other data I’ve been privy to, sales numbers on the Oculus store may be about 50% of sales on Steam. I have no idea how reliable this number is, or what the variance is, but it at least provides a starting point for ballpark estimates.

Rough player counts are possible for Steam through Steamspy, and Playstation through Gamstat, but ultimately without access to the raw data behind each of these platforms, opportunities for quantitative analysis are limited (as are my skills in that regard). However, obviously some patterns have emerged.

It seems to me that this general alignment between what people are continuing to play and what is selling well is a sign that 2018 was the first year of stability in the VR market. Games can no longer benefit easily from a first mover’s advantage, where people will buy a game simply because it fills a gap in the market.

Below are the best-selling VR games in 2018, along with ownership numbers. The games were sorted into tiers based on the sales achieved in 2018, meaning some games in lower tiers have higher owner counts than games in higher tiers, due to release date or sales pattern differences.

Platinum (Tier 1)

Game Release User Score Owners
Beat Saber 2018 [EA] 97% 563,000
Pavlov VR 2017 [EA] 89% 370,000
H3 VR 2016 [EA] 97% 297,000
Job Simulator 2016 84% 280,000
SUPERHOT VR 2017 89% 262,000
Onward 2016 [EA] 91% 256,000
Arizona Sunshine 2016 86% 249,000
Skyrim VR 2018 82% 215,000
Fallout 4 VR 2017 71% 201,000
GORN 2017 [EA] 97% 195,000
OrbusVR 2017 [EA] 81% 29,000

Gold (Tier 2)

Game Release User Score Owners
Space Pirate Trainer 2017 [prev. EA] 95% 164,000
DOOM VFR 2017 59% 119,000
Raw Data 2017 [prev. EA] 87% 97,000
Rick and Morty 2017 74% 89,000
I Expect You To Die 2017 92% 68,000
Budget Cuts 2018 73% 49,000
STAND OUT 2017 [EA] 77% 44,000
Zero Caliber VR 2018 [EA] 73% 39,000
Sairento VR 2018 [prev. EA] 90% 38,000
Sprint Vector 2018 90% 28,000

Silver (Tier 3)

Game Release User Score Owners
Audioshield 2016 82% 126,000
Serious Sam: The Last Hope 2017 85% 79,000
Blade & Sorcery 2018 86% 74,000
Fruit Ninja VR 2016 85% 69,000
Dead Effect 2 VR 2017 83% 55,000
Richie’s Plank Experience 2017 83% 45,000
Moss 2018 91% 43,000
VTOL VR 2017 94% 37,000
In Death 2018 91% 29,000
Duck Season 2017 86% 28,000
Creed 2018 82% 27,000
Talos Principle VR 2017 87% 23,000
Box VR 2017 87% 22,000
Serious Sam 3 VR 2017 89% 19,000
LA Noire 2017 62% 18,000

Filed under Analysis, Blog, Games Tagged with , , , , ,

A Manifested SLS

April 26, 2017 Leave a comment

NASA recently released details on their planned architecture for human spaceflight in the coming two decades. This was a welcome update, because before now the plan was little more than “we build SLS, we do something around the Moon, then we go to Mars”. You may think that I’m joking or exaggerating. I’m not.

Fortunately, we don’t have to speculate in the dark anymore. We have hard details about what the SLS will be launching and how it will help NASA eventually land humans on Mars. Well, sort of. We have details about what the SLS will be launching, in any case.

NASA’s Plan

The plan revolves around two new pieces of infrastructure: the Deep Space Gateway (DSG) and the Deep Space Transport (DST). The DSG is essentially a space station in orbit around the Moon. The DST is a spacecraft for transporting humans from the DSG to Mars. The DST is reusable, and only stays in orbit. That is, it is simply a shuttle between Martian orbit and lunar orbit. Both the DSG and DST will be equipped with solar electric propulsion (SEP) modules, allowing the DSG to change its orbit around the Moon and allowing the DST to fly between the Moon and Mars without ditching any mass or refueling (electric propulsion is very efficient). Finally, the Orion capsule is used to transport astronauts between Earth and the DSG.

The SLS, NASA’s big new rocket, can launch an Orion capsule to lunar orbit along with 10 metrics tons (mT) of cargo (this configuration is called a co-manifested payload, or CMP). The SLS can also launch 40 mT of cargo to the Moon if launched without a crew. The DSG will be composed of four modular chunks, each launched as a CMP with a group of astronauts (one launch per year from 2022 to 2025). The astronauts will help assemble the DSG in lunar orbit before returning to Earth.

After the DSG is complete, a cargo launch of the SLS will loft the DST in 2027 as a monolithic (non-modular) spacecraft. Astronauts will run the DST through a shakedown cruise around the Moon in 2029, before embarking on some sort of Mars mission after 2030.

You might notice that this plan doesn’t actually specify what the Mars missions will look like. There are some vague hand-waved mentions of a mission to Phobos or Deimos (the moons of Mars), and then maybe landing on Mars. What kind of science will the DST be able to do? Surface samples from Phobos? In-situ resource utilization (ISRU) experiments? How will NASA send landers to Mars? What will they look like? Will NASA be landing support equipment on the surface ahead of time? Instead of going to Mars quickly and cheaply, we are going to spend a decade or more puttering around in lunar orbit, and it’s not clear why. At the end of this (unintentionally long) post, I will propose a plan that accomplishes the same thing with less money, less time, and less infrastructure.

Why?

But before addressing these more concrete questions, there are some existential questions that need to be addressed. For example: why? Why are we spending billions of dollars to launch a handful of humans into lunar orbit, and why are they monkeying around up there?

In an effort to answer these questions, let’s step back and answer the larger question: why do we care about sending people into space? There are certainly motivations for having infrastructure in space: GPS, communications, Earth and space weather monitoring, climate research. These activities have real economic value, but they don’t justify sending humans.

There is the scientific motivation: learning more about the stars and about the formation of the solar system and the origin of life. This is the rationale that NASA often cites. However, there are countless arguments to be made in favor of widespread robotic exploration: it’s cheaper and faster, robots can go more places, robots don’t have to come back, robots can stick around for longer, we aren’t risking planetary contamination, etc. These concerns outweigh the few benefits that humans do provide.

A third motivation, one that I suspect a lot of people at NASA actually believe in but would never dare say: we landed people on the Moon, then we stopped. We need to land people on Mars because space travel is cool, but we can’t justify just going back to the Moon. I think this stems from a larger infatuation with the vision of the future instilled in us by decades of media consumption. Science fiction depicts humanity among the stars, so we must go. I suspect this motivation is rather rampant among human spaceflight fans. Needless to say, this motivation holds little value.

The only meaningful motivator for human spaceflight is neither economic nor scientific. The only reason to kick off interplanetary human spaceflight (and therefore the only reason to be interested in human spaceflight at all) is the distant goal of establishing a permanent, self-sustaining human presence off of the Earth in order to make sure our species never dies out. As a species, we must invest in the intrinsic value of this motivator, or give up human spaceflight once and for all. Doing it for any other reason is simply masturbatory.

Ok, so we care about sending people into space because we eventually want to establish a self-sufficient colony. We care about going to Mars because the surface of Mars is, for a whole host of reasons, the best place to create that colony. But why do we care about spending billions of dollars on an unscalable architecture to send a handful of humans on short trips around the Moon and Mars?

The answer is: maybe we shouldn’t. There are good, well-formed arguments against the whole endeavor. However, NASA has always been an organization that pioneers research into really difficult problems, and then gives their results to the public so that private companies can improve and capitalize on those advancements. If the government is set on spending billions of dollars on human spaceflight, we might as well spend it in a way that will help the eventual colonization of Mars.

So what kinds of unanswered questions or nebulous roadblocks stand between humanity and our intrinsically-valued goal of surviving a potential devastation of the Earth? Cutting past the leagues of logistical questions about a surface colony:

  • Data on the effect of long-term partial gravity on human biology
  • More data on the effect of long-term (500+ days) exposure to solar radiation
  • More data on the psychological effect of long-term separation from Earth
  • Research on high-efficiency life support systems
  • Research on long-duration missions without access to resupply missions from Earth
  • A platform for small companies to test technologies from various staging points around the solar system (more on this later)

These can be divided into two categories: research on human factors, and research on vehicle systems which until now have been unnecessary.

Fortunately, thanks to Mir and the ISS, we have lots of data on the effects of long-term weightlessness on human biology, and lots of data on how space vehicles degrade over the course of years. However, we still lack experience with maintaining space vehicles without constant access to the Earth (e.g. manufacturing new parts in space or creating highly redundant systems). We also don’t have experience creating closed-loop life support systems with extremely efficient recycling (the ISS is rather wasteful).

We can also gather a lot of data about the human factors on Earth. Data on radiation exposure from nuclear workers and people exposed to radiation sources is abundant. We can run more experiments like Mars 500 to gather psychological data. Researching the effects of long-term partial gravity is unfortunately quite difficult, but it is also one of the least important factors for now. We can safely ignore it.

Finally, it must be considered a necessity to include commercial efforts at every step in the process. Ultimately, it will be corporate entities that enable and support colonization efforts. NASA can and should play a critical role in enabling the development and testing of technologies by private entities. We have seen how valuable this synergy can be in the ISS; from launching cubesats to mounting experimental modules, smaller companies can gain invaluable operational experience. Resupply contracts from NASA essentially kept SpaceX afloat, arguably one of NASA’s more important contributions in the last 20 years.

So, any effort by NASA should focus on two goals: tackle the problems of closed-loop life support and long-duration mission maintenance, and provide a configurable, expandable framework for commercial involvement. Naturally, this necessitates the launching of astronauts beyond Earth orbit, so I think we’ve solved our existential question. Let’s move on to critiquing NASA’s proposal.

Being Useful

NASA has done a fantastic job of shoehorning itself into a host of arbitrary design constraints by jumping from plan to plan over the course of 4 administrations. Let’s enumerate them, and then evaluate the new proposal by its satisfaction of the constraints and goals.

  • The Orion MPCV exists. It can only be launched by the SLS, it can carry 4 people, and it can only re-enter from lunar velocities or less (why didn’t they design it for re-entry from Martian velocities, you ask? Good question).
  • The SLS exists. Block 1B can carry 105 mT to LEO, and 39 mT to a Translunar Injection (TLI). If it launches a co-manifested payload (CMP) with Orion, it can send 10 mT to TLI. Block 2 (first flight planned 2029) can carry 130 mT to LEO and 45 mT to TLI. It will fly once a year.
  • We want to utilize the 40 kW SEP module that was designed for the cancelled Asteroid Redirect Mission (ARM).
  • We have to do things around the Moon. Because reasons.

That last bullet point is motivated by the annoyingly persistent faction within NASA and the general spaceflight community that advocates for the development of a lunar infrastructure. I could write a whole other blog post on why any effort to develop infrastructure on the lunar surface is a huge waste of time, but needless to say, we should avoid it.

Let’s look at the decision to recycle ARM technology to propel the DSG and DST. There are a couple of good arguments for using SEP to reach Mars. Since the thrusters are much more well-behaved than chemical propulsion, we don’t have to worry about losing engines (which would spell certain doom for the mission). You also don’t have to worry about the boil-off of cryogenic propellants during transit. Additionally, your specific impulse is much higher, meaning the propellant mass can be lower. Or, alternatively, with the added delta-V from the same fuel mass, we can perform more orbital maneuvers or even return from Mars without ditching any parts of the vehicle.

There are also some arguments against SEP. It is slow to accelerate, which means your astronauts are going to be spending more time in space (bad). It also means that you have to stage the mission from Lunar orbit or a Lagrange point, since the vehicle would take multiple years to reach escape velocity from low Earth orbit (LEO). Staging the mission from beyond LEO means mass budgets are tighter. You also can’t refuel the vehicle using ISRU, ever. Since the propellant is argon or xenon, rather than methalox or hydrolox, you can’t refine it from water and carbon dioxide.

Given these tradeoffs, it’s easy to see why NASA would pick SEP for its long-duration human spaceflight missions. NASA is risk averse: they don’t want to rely on ISRU (a huge technical and logistical unknown) for a critical mission component, so that downside of SEP doesn’t matter to them. NASA detests the idea of resting mission success on machinery as notoriously unreliable and complicated (read: unrepairable) as a chemical thruster. Finally, the fact that SEP “forces” NASA to develop a lunar infrastructure is just icing on the cake.

The fact that using SEP means you can get the whole vehicle back from Mars rather shakes up the design of a Mars mission architecture. It suddenly makes sense to assemble your mission in cis-lunar space rather than launching each monolithic component directly to Mars, like in the Mars Direct architecture. And if you are assembling a mission in cis-lunar space, then it makes sense to build a space station to assist you logistically. If your mission has to be assembled in lunar orbit (because of SEP’s low thrust) then you have to build the space station in lunar orbit, which means you need a large rocket capable of lofting space station components into lunar orbit. Suddenly everything NASA has been doing, from the ARM to the SLS, becomes rationalized.

Except, wait a second. This sounds suspiciously like the original proposal for the STS. You build a space shuttle to launch space station and spacecraft components, and then you send out missions to the Moon and Mars from LEO. In the case of the STS, the Moon and Mars missions got cancelled, leaving only the space shuttle and a pared-down space station. Suddenly you have a space station that doesn’t serve any purpose beyond providing justification for continuing the shuttle program.

The new proposal falls along the same lines so perfectly that I suspect it may have been intentional. The first launch of a DSG segment is in 2022, which (if Trump gets re-elected) will be near the end of this administration. With a piece of hardware in space, the next administration will have no choice but to continue its construction. The DST will get cancelled or transformed into something else, and NASA will have an excuse to keep launching the SLS, and the DSG will become the next ISS (as the ISS is decommissioned in 2024, around the first time a crew is launched to the DSG).

Regardless of politics, the proposal meets all the constraints. The Orion is regularly sent to lunar orbit carrying astronauts and a 10 mT CMP to be added to a space station around the Moon. The astronauts spend their time assembling the space station and testing out its thruster module, which is a modified version of the propulsion module from the ARM. You use the heavy-lift capacity of the SLS to launch a SEP spacecraft to rendezvous with the station in lunar orbit, then send astronauts to test it out before making excursions to Mars. It’s so neat and tidy that I could put a bow on it.

It’s so neat that it’s easy to forget that the architecture barely addresses any of the useful existential reasons for sending humans into space in the first place. With annual resupply missions to the DSG and no permanent human habitation, there is little motivation to develop a closed-loop life support system. This only happens when we get to the DST. The strange thing, then, is that the shakedown cruise of the DST happens while attached to the DSG. A mission which is supposed to demonstrate that the DST could operate for three years without resupply or repair instead gets access to 40 mT of equipment and logistical supplies in the DSG—something that won’t be available during the DST’s real cruise to Mars. There is some room for research on long-duration mission maintenance with the DST, although it seems remote from NASA’s goals or desires.

At least the proposal does a good job of providing a framework for commercial involvement. There are even explicit slots on the manifest for commercially-contracted launches to service the DSG, and one can imagine that the station could play a support role for any commercial missions on the surface of the Moon (e.g. teleoperation and observation of the surface from above). The DSG could play a similar role as the ISS, deploying small scientific and commercial payloads into various lunar orbits (given its ability to perform extensive orbital maneuvers), and playing host to commercially-constructed modules.

The proposal doesn’t do a great job of enabling the exploration of Mars, unfortunately. The key problem, I think, is that NASA doesn’t plan on making more than a single DST. This is confusing to me, because it means that heavy payloads like landers, rovers, and ISRU experiments will need to be launched and transported separately. If you had multiple DSTs acting as tugs for both habitation modules and support equipment, you could assemble a mission in cis-lunar space at the DSG. As it is, any missions that need serious support hardware will have to rendezvous in Martian orbit.

So this means that the only purpose of the DSG is to provide a rendezvous point for the DST and the Orion capsule. In other words, the DSG doesn’t serve any purpose at all. Remember that the DSG only made sense as a logistical support for the assembly of a Mars mission in lunar space. If there isn’t any assembly required, you don’t need a rendezvous point.

Let’s break it down: you launch in an Orion and rendezvous in lunar orbit. You transfer to the DST, take it to Mars, and then come back. You transfer to an Orion capsule, and return to Earth. At no point does the DSG play a role beyond “crew transfer tube”. If you think it’s a place to store your Orion until you get back, think again; the Orion capsule is designed to support crew actively for no more than 21 days, and to stay in space no more than 6 months. That means that the capsule you are returning to Earth in is different from the capsule you launched in. At some point, an Orion will have to make an automated rendezvous in lunar orbit, and an Orion will have to make an unmanned return from lunar orbit. This renders the DSG rather useless as a component of a Mars mission architecture.

There are more reasons why only building a single DST doesn’t make any sense. First of all, putting crew in an untested vehicle is anathema to NASA—wouldn’t it be smart to send an uncrewed DST on a shakedown cruise, perhaps even using it to boost a scientific payload to Mars (hint: Phobos sample return mission)? Then they can iterate on the design and start building crewed DSTs. In the current plan, the lone DST gets a crewed shakedown cruise. What happens when they discover a problem? Do they try to jury-rig a fix? Do they ship replacement parts from Earth? Do they abandon the mission? Since NASA won’t be building future DSTs, there aren’t any opportunities to employ the hard-learned lessons from operating a space vehicle for a long duration. This defeats the whole purpose of NASA doing a Mars mission in the first place.

I don’t think NASA has put any real stock in this plan. Why would NASA launch the DSG as four 10mT chunks riding along with an Orion each time, when the same station could be launched as a single 40mT monolithic block by a single cargo launch? Why even build the DSG? Why build only one DST?

The Real Plan

Here’s what I think. This whole mission architecture is a sly way of fulfilling NASA’s dream of returning to the Moon and reliving the glory days of Apollo. The political likelihood of the DST getting cancelled in 2024 is high, given a turnover of the administration and the fact that the plan doesn’t do a great job of getting us to Mars. With the DST cancelled but the production of the remaining components of the DSG in high-gear, the new administration will be forced to authorize the completion of the DSG to avoid looking bad. However, to distance themselves from the previous administration, they will mandate that instead of being used for future Mars missions, the DSG will play a critical support role for future Moon missions.

This makes a lot of sense, because the DSG is fantastic if your plan is to send people to the Moon, rather than to Mars. You can dock a single-stage lunar lander to the DSG and send refueling missions from Earth between each sortie to the surface. Moreover, NASA can fund companies interested in mining water on the Moon because the DSG will need a steady supply of water and oxygen for its astronauts. From 2024 to 2032, NASA will be positioned to enable a veritable renaissance for lunar exploration and exploitation.

I will leave it to a future blog post to explain why focusing on the Moon is a huge waste of humanity’s time and America’s money. For now, let it suffice to say that it doesn’t bring us any closer to our original goal of aiding the eventual settlement of a self-sustaining colony on Mars.

Assuming that what NASA has proposed is notionally feasible, I propose that with a little bit of reordering and restructuring, the plan can be turned into something that actually advances humanity towards our distant goal of colonization.

My Plan

First, NASA’s plan calls for an uncrewed test launch of the SLS and Orion in 2019, which is a good idea in my book. It tests out the SLS, and let’s NASA test the Orion capsule in a multi-week mission in lunar orbit. However, the mission is slated to use a temporary second stage, because the so-called Exploration Upper Stage (which all future SLS missions will use) will not be ready by 2019. NASA’s plan also calls for launching a probe to Jupiter in 2021, which I also think is a great idea. It allows NASA to test out the new second stage without crew on-board.

However, with the 2022 launch of the SLS, I propose that the entire DSG is launched as a monolithic 40 mT station. A crewed launch of the SLS in the following year would spend up to 6 months aboard the station, consuming the 10 mT of food, air, and water brought along as a CMP. This would give them a chance to test all the station’s system, perform any necessary setup or repair tasks, maneuver the station into various orbits, and release some cubesats. Much like Skylab, there will likely be some issues discovered after the station’s launch, giving NASA a year to develop some fixes and include them in the 10 mT of CMP cargo.

One might object by pointing out that this means NASA will have to construct the logistics, habitation, and airlock modules of the DSG up to three years sooner than in the initial plan, and accelerated timelines are a Bad Thing. I would counter by pointing out that by making the station monolithic, all the mass required for docking interfaces and independent power management is removed. The station would also be much less volume-constrained, as they would only be constrained by the large 8.4 meter cargo fairing, rather than the significantly smaller interstage fairing that a CMP must fit into. This means that NASA engineers wouldn’t need to spend as long shaving off mass and fitting components into a smaller volume. NASA would also have a year or two of schedule wiggle room before the next presidential election and change of administration.

After the first crewed SLS launch in 2023 to the monolithic DSG, I propose a year gap in the schedule. I find it unlikely that they can ramp up from a first launch in 2019 to a launch every year in 2021, 2022, and 2023. There will be schedule slip. So the next crew launch would be in 2025 (potentially earlier, if they can manage it), bringing a 10 mT life support module as CMP to test new closed-loop life support technology. The crew would stay this time for roughly a year, testing the new life support capabilities and receiving shipments of commercial equipment.

To that end, starting in 2024 or earlier, commercially-contracted missions would resupply the DSG with life support consumables and fuel. They would also haul up new modules and equipment, ranging from expandable habitat modules to scientific sensors and communications arrays for controlling surface rovers. These launches would ideally be timed to occur before or during a crewed mission. While the uncrewed freighters would likely be able to dock autonomously, the crew would be required for installing new equipment both inside and outside the station.

In 2026, a prototype DST would be launched with a probe for taking samples from Phobos or Deimos. It would leave for Mars in the 2026 transfer window, and leave Mars in 2028 to return in July 2029. A cargo launch in 2027 would launch a second DST equipped with an integrated habitation and life support module, which would dock with the DSG and await the next crew launch.

A crewed mission launched in early 2028 or late 2027 would remain aboard the DSG until the prototype DST returned from its mission (staying for more than a year, as a dress rehearsal for a flight to Mars). The astronauts would perform a checkout of the returned DST and return to Earth with Phobos/Deimos soil samples from the probe. During their stay, they could perform a shakedown flight of the second DST in lunar space. After their departure, the first DST could be repurposed for a second unmanned mission or (more likely) put through rigorous operations in lunar orbit for stress testing.

Finally, two launches in late 2028 and early 2029 would loft a third DST with a small Martian lander equipped with ISRU, and a crew to transfer into the DST that launched in 2027. Both DSTs would depart immediately for Mars during the transfer window. The lander would carry out ISRU experiments on the surface while the crew remotely operated it from orbit. This presents a nice opportunity to allow sample retrieval from a previous sample-collecting rover mission, launching the sample into orbit using an ascent craft carried down by the ISRU lander. The astronauts in Martian orbit would retrieve the ascent craft and return to Earth with Mars surface samples. They would depart during the 2030 window, returning in September 2031. This mission would last for more than a year and a half, setting the stage for more aggressive missions in the future.

Let’s compare my plan against NASA’s: the DSG is operational by 2022 in my plan, rather than 2026. As a downside, the first crewed flight occurs a year later in my plan, and only two crewed missions take place by 2026, as opposed to NASA’s four. However, in my plan the second mission would last for about a year, rather than NASA’s 16-42 day missions. Under my plan, the DST would launch a year earlier and immediately be subjected to a test flight to Mars. Also, my plan piggybacks a valuable scientific probe on this test mission, allowing for an unprecedented scientific return (nobody has ever returned samples from other moons or planets).

When NASA’s plan has a crew performing a 221-day checkout mission for the first time in 2027, my plan calls for a roughly 400-day endurance mission around the same time that both runs diagnostics on a crewed DST (the purpose of the 221-day flight in NASA’s plan) and returns with samples from Phobos or Deimos. When NASA’s plan has astronauts performing a 400-day shakedown mission on the DST around the Moon (2029), my plan has astronauts on their way to Mars, with multiple long-endurance missions under their belt to validate long-duration survival in deep space.

Best of all, NASA’s plan uses 12 SLS launches before getting around to a crewed Mars mission “sometime after 2030”, while my plan uses 10 SLS launches to achieve a crewed Mars mission and performs two sample return missions to boot.

I think the most reasonable explanation, as I iterated above, is that NASA is trying to put off its Mars missions as long as possible until a new presidential administration redirects them to focus on lunar activities alone. I will leave it to another blog post to debunk the rationalizations for this ambition.

Filed under Analysis, Blog, Space Tagged with , , , , , , ,

The Interplanetary Transport System

October 21, 2016 3 Comments

Space has been on my brain a lot lately. One of the causes was the long-awaited presentation by Elon Musk at the International Astronautical Congress (IAC) last month. During the talk, he finally laid out the details of his “Interplanetary Transport System” (ITS). The architecture is designed to enable a massive number of flights to Mars for absurdly low costs, hopefully enabling the rapid and sustainable colonization of Mars. The motivation behind the plan is a good one: humanity needs to become a multi-planetary species. The sheer number of things that could take civilization down a few pegs or destroy it outright is frighteningly lengthy: engineered bio-weapons, nuclear bombs, asteroid strikes, and solar storms crippling our electrical infrastructure are some of the most obvious. Rampant AI, out-of-control self-replicating robots, and plain old nation-state collapse from war, disease, and famine are some other threats. In the face of all those horrifying things, what really keeps me up at night is the fact that if civilization collapses right now, we probably won’t get another shot. Ever. We’ve abused and exhausted the Earth’s resources so severely, we simply cannot reboot human civilization to its current state. This is the last and best chance we’ll ever get. If we don’t establish an independent, self-sufficient colony on Mars within 50 years, we’ll have solved the Fermi Paradox (so to speak).

But Musk’s Mars architecture, like most of his plans, is ambitious to the point of absurdity. It at once seems like both fanciful science fiction and impending reality. Because Musk works from first principles, his plans defy socio-political norms and cut straight to the heart of the matter and this lateral approach tends to rub the established thinkers of an industry the wrong way. But we’ve seen Musk prove people wrong again and again with SpaceX and Tesla. SpaceX has broken a lot of ground by being the first private company to achieve orbit (as well as return safely to Earth), to dock with the International Space Station, and to propulsively land a part of a rocket from an orbital launch. That last one is particularly important, since it was sheer engineering bravado that allowed them to stand in the face of ridicule from established aerospace figureheads. SpaceX is going to need that same sort of moxie in spades if they are going to succeed at building the ITS. Despite their track record, the ITS will be deceptively difficult to develop, and I wanted to explore the new and unsolved challenges that SpaceX will have to overcome if they want to follow through on Musk’s designs.

SpaceX ITS Diagram

The basics of the ITS architecture are simple enough: a large first stage launches a spaceship capable of carrying 100 people to orbit. More spaceships (outfitted as tankers) are launched to refill the craft with propellants before it departs for Mars during an open transfer window. After a 3 to 6 month flight to the Red Planet, the spaceship lands on Mars. It does so by at first bleeding off speed with a Space Shuttle-style belly-first descent, before flipping over and igniting its engines at supersonic speeds for a propulsive landing. After landing, the craft refill its tanks by processing water and carbon dioxide present in Mars’s environment and turning them into propellant for the trip back to Earth. Then the spaceship simply takes off from Mars, returns to Earth, and lands propulsively back at home.

Now, there are a lot of hidden challenges and unanswered questions present in this plan. The first stage is supposed to land back on the launch mount (instead of landing on a pad like the current Falcon 9 first stage), requiring centimeter-scale targeting precision. The spaceship needs to support 100 people during the flight over, and the psychology of a group that size in a confined space for 6 months is basically unstudied. Besides other concerns like storing highly cryogenic propellants for a months-long flight, radiation exposure during the flight, the difficulty of re-orienting 180 degrees during re-entry, and the feasibility of landing a multi-ton vehicle on soft Martian regolith using powerful rocket engines alone, there are the big questions of exactly how the colonists will live and what they will do when they get to Mars, where the colony infrastructure will come from, how easy it will be to mine water on Mars, and how the venture will become economically and technologically self-sufficient. Despite all of these roadblocks and question marks, the truly shocking thing about the proposal is the price tag. Musk wants the scalability of the ITS to eventually drive the per-person cost down to $200,000. While still high, this figure is a drop in the bucket compared to the per-capita cost of any other Mars architecture on the table. It’s well within the net-worth of the average American (although that figure is deceptive; the median American net-worth is only $45,000. As far as I can figure, somewhere between 30% and 40% of Americans would be able to afford the trip by liquidating most or all of their worldly assets). Can SpaceX actually achieve such a low operational cost?

Falcon 9 Production Floor

Remember that SpaceX was originally targeting a per-flight price of $27 million for the Falcon 9. Today, the price is more like $65 million. Granted, the cost to SpaceX might be more like $35 million per flight, and they haven’t even started re-using first stages. But it is not a guarantee that SpaceX can get the costs as low as they want. We have little data on the difficulty of re-using cores. Despite recovering several in various stages of post-flight damage, SpaceX has yet to re-fly one of them (hopefully that will change later this year or early next year).

That isn’t the whole story, though. The Falcon 9 was designed to have the lowest possible construction costs. The Merlin engines that power it use a well-studied engine design (gas generator), low chamber pressures, an easier propellant choice (RP-1 and LOX), and relatively simple fabrication techniques. The Falcon 9 uses aluminum tanks with a small diameter to enable easy transport. All of their design choices enabled SpaceX to undercut existing prices in the space launch industry.

But the ITS is going to be a whole other beast. They are using carbon fiber tanks to reduce weight, but have no experience in building large (12 meter diameter) carbon fiber tanks capable of holding extremely cryogenic liquids. The Raptor engine uses a hitherto unflown propellant combination (liquid methane and liquid oxygen). Its chamber pressure is going to be the highest of any engine ever built (30 MPa. The next highest is the RD-191 at 25 MPa). This means it will be very efficient, but also incredibly difficult to build and maintain. Since reliability and reusability are crucial for the ITS architecture, SpaceX is between a rock and a hard place with its proposed design. They need the efficiency to make the system feasible, but the high performance envelope means the system will suffer less abuse before needing repairs, reducing the reusability of the system and driving up costs. At the same time, reusability is crucial because the ITS will cost a lot to build, with its carbon fiber hull and exacting standards needed to survive re-entry at Mars and Earth many times over.

It’s almost like the ITS and Falcon 9 are on opposites. The Falcon 9 was designed to be cheap and easy to build, allowing it to be economical as an expendable launch vehicle, while still being able to function in a large performance envelope and take a beating before needing refurbishment. The ITS, on the other, needs all the performance gains it can get, uses exotic materials and construction techniques, and has to be used many times over to make it an economical vehicle.

All of these differences make me think that the timeline for the development of the ITS is, to put it mildly, optimistic. The Falcon 9 went from the drawing board to full-stack tests in 6 years, with a first flight a few years later. Although the SpaceX of 2004 is not the SpaceX of 2016, the ITS sure as hell isn’t the Falcon 9. A rocket using the some of the most traditional and well-worn engineering methods in the book took 6 years to design and build. A rocket of unprecedented scale, designed for an unprecedented mission profile, using cutting-edge construction techniques… is not going to take 6 years to design and build. Period. Given SpaceX’s endemic delays with the development of the Dragon 2 and the Falcon Heavy, which are a relatively normal sized spaceship and rocket, respectively, I suspect the development of a huge spaceship and rocket will take more like 10 years. Even when they do finally fly it, it will take years before the price of seat on a flight falls anywhere as low as $200,000.

Red Dragon over Mars

If SpaceX manages to launch their Red Dragon mission in time for the 2018 transfer window, then I will have a little more hope. The Red Dragon mission needs both a proven Falcon Heavy and a completely developed Dragon 2. It will also allow SpaceX to answer a variety of open questions about the mission profile of the ITS. How hard is it to land a multi-ton vehicle on Martian regolith using only a powered, propulsive descent? How difficult will it be to harvest water on Mars, and produce cryogenic propellants from in situ water and carbon dioxide? However, if SpaceX misses the launch window, I definitely won’t be holding my breath for humans on Mars by 2025.

Filed under Analysis, Blog, Space Tagged with , , , , ,

What Does It Take To Become A Programmer?

June 5, 2016 3 Comments

So these are my thoughts on this article (hint, it’s utter tripe): Programming Doesn’t Require Talent or Even Passion.

On the one hand, this article espouses a good sentiment (you don’t have to be gifted to learn programming). On the other, it completely disregards the important idea that being able to do something is not the same as being able to do it well.

I can draw, but anyone who has seen me draw would agree that I’m pretty bad at it. I can draw just well enough to get my concepts across to other people. However, if I intended on becoming an artist for a living, I should probably learn about proportions, shading, composition, perspective, color theory, and be able to work with a range of mediums. Of course, there isn’t some big secret to learning these things. You just practice every day and study good artistic work, analyzing how it was made. Maybe you take some courses, or read some books that formally teach certain techniques. After thousands of invested hours, you will find that your drawing has radically improved, as shown again and again by progress comparison pictures (that one is after 2 years of practice).

The same holds true for programming. Anyone can learn programming. It requires nothing except a little dedication and time. But the article starts out by promising to ‘debunk’ the following quote (I’m not sure if it’s actually a real quote – they don’t attribute it to anybody):

You not only need to have talent, you also need to be passionate to be able to become a good programmer.

The article immediately ignores the fact that the ‘quote’ is talking about good programmers. Just like becoming a good artist requires artistic talent and a passion for learning and improving every day, good programmers are driven by the need to learn and improve their skills. Perhaps an argument can be made for “talent” being something you acquire as a result of practice, and thus you don’t need talent to start becoming good; you become good as you acquire more and more talent. This is a debate for the ages, but I would say that almost invariably a passion for a skill will result in an early baseline proficiency, which is often called “talent”. Innate talent may or may not exist, and it may or may not influence learning ability.

It doesn’t really matter though, because the article then goes on to equate “talent” and “passion” with being a genius. It constructs a strawman who has always known how to program and has never been ignorant about a single thing. This strawman, allegedly, causes severe anxiety to every other programmer, forcing them to study programming at the exclusion of all else. It quotes the creator of Django (after affirming that, yes, programmers also suffer from imposter syndrome):

Programming is just a bunch of skills that can be learned, it doesn’t require that much talent, and it’s not shameful to be a mediocre programmer.

Honestly, though, the fact of the matter is that being a good programmer is incredibly valuable. If your job is to write code, you should be able to do it well. You should write code that doesn’t waste other people’s time, that doesn’t break, that is maintainable and performant. You need to be proud of your craft. Of course, not every writer or musician or carpenter takes pride in their craft. We call these people hacks and they churn out deplorable fiction that only shallow people read, or uninteresting music, or houses that fall down in an earthquake and kill dozens of people.

So, unless you want to be responsible for incredibly costly and embarrassing software failures, you better be interested in becoming a good programmer if you plan on doing it for a career. But nobody starts out as a good programmer. People learn to be good programmers by having a passion for the craft, and by wanting to improve. If I look at older programmers and feel inferior by comparison, I know it’s not because they are a genius while I am only a humble human being. Their skill is a result of decades of self-improvement and experience creating software both good and bad.

I think it’s telling that the article only quotes programmers from web development. Web development is notorious for herds of code monkeys jumping from buzzword to buzzword, churning out code with barely-acceptable performance and immense technical debt. Each developer quote is followed by a paragraph that tears down the strawman that was erected earlier. At this point, the author has you cheering against the supposedly omnipresent and overpowering myth of the genius programmer — which, I might remind you, is much like the myth of the genius painter or genius writer; perhaps accepted by those with a fixed mindset, but dismissed by anybody with knowledge of how the craft functions. This sort of skill smokescreen is probably just a natural product of human behavior. In any case, it isn’t any stronger for programming than for art, writing, dance, or stunt-car driving.

The article really takes a turn for the worse in the second half, however. First, it effectively counters itself by quoting jokes from famous developers that prove the “genius programmer” myth doesn’t exist:

* One man’s crappy software is another man’s full time job. (Jessica Gaston)

* Any fool can write code that a computer can understand. Good programmers write code that humans can understand.

* Software and cathedrals are much the same — first we build them, then we pray. (Sam Redwine)

The author LITERALLY ASKS: “If programmers all really had so much talent and passion, then why are these jokes so popular amongst programmers?”, as if to prove that he was being intellectually dishonest when he said back in the beginning “It’s as if people who write code had already decided that they were going to write code in the future by the time they were kids.”

But the absolute worst transgression the article makes is quoting Rasmus Lerdorf, creator of PHP. PHP is a server-side language. It is also one of the worst affronts to good software design in recent history. The reason it was the de facto server-side language before the recent Javascript explosion is that it can be readily picked up by people who don’t know what they are doing. Like you would expect from a language designed by someone who “hates programming” and used by people who don’t what they are doing, PHP is responsible for thousands of insecure, slow, buggy websites.

PHP’s shortcoming are amusingly enumerated in this famous post: PHP – a fractal of bad design. In the post, the following analogy is used to illustrate how PHP is bad:

I can’t even say what’s wrong with PHP, because— okay. Imagine you have uh, a toolbox. A set of tools. Looks okay, standard stuff in there.

You pull out a screwdriver, and you see it’s one of those weird tri-headed things. Okay, well, that’s not very useful to you, but you guess it comes in handy sometimes.

You pull out the hammer, but to your dismay, it has the claw part on both sides. Still serviceable though, I mean, you can hit nails with the middle of the head holding it sideways.

You pull out the pliers, but they don’t have those serrated surfaces; it’s flat and smooth. That’s less useful, but it still turns bolts well enough, so whatever.

And on you go. Everything in the box is kind of weird and quirky, but maybe not enough to make it completely worthless. And there’s no clear problem with the set as a whole; it still has all the tools.

Now imagine you meet millions of carpenters using this toolbox who tell you “well hey what’s the problem with these tools? They’re all I’ve ever used and they work fine!” And the carpenters show you the houses they’ve built, where every room is a pentagon and the roof is upside-down. And you knock on the front door and it just collapses inwards and they all yell at you for breaking their door.

That’s what’s wrong with PHP.

And according to Rasmus Lerdorf, the creator of this language:

I’m not a real programmer. I throw together things until it works then I move on. The real programmers will say “Yeah it works but you’re leaking memory everywhere. Perhaps we should fix that.” I’ll just restart Apache every 10 requests.

It’s like the article is admitting that if you don’t take the time to learn good programming principles, you are going to be responsible for horrible systems that cause headaches five years down the line for the people maintaining them and that regularly allow hackers to access confidential personal information like patient information and social security numbers for millions of people.

So yes, if you aren’t planning on programming for a career, learning to program is fairly straightforward. It’s as easy as learning carpentry or glass-blowing. It might seem daunting, but invest a half dozen hours and you can have your foot solidly in the door.

But if you plan on building systems other people will rely on, you sure are hell better pick up some solid programming fundamentals. If you aren’t motivated to improve your skillset and become a better programmer, don’t bother learning at all. Don’t be the reason that the mobile web sucks, and don’t be the reason that 28 American soldiers died. Learn to be a good programmer.

Filed under Analysis, Blog Tagged with , , , , , ,

Indiscriminately Valuing Non-Violent Games

May 1, 2016 Leave a comment

Starting with the 1980s arcade games Galaxian and Missile Command, games and combat became nearly synonymous. This was only exacerbated in the 90s by the advent of wildly popular shooters like Doom. The choice to focus a game around antagonism, combat, and violence was not a conscious design decision, but a necessity of the industry and environment. There were abstract games that didn’t contain violence, but in general the highest-profile games were about, in essence, murder.

Doom screenshot

Doom: you shoot things. Dead simple.



Then a renaissance occurred in academia, and suddenly games were art. Nobody really knew what to do with this fact or what it meant, but it was revolutionary, and regardless of anything else, games were definitely art. To support this, a number of innovative (perhaps iconoclastic) non-violent games — games like Journey and Gone Home — were foisted up as evidence that games are art. “Games are art, they can convey aesthetics beyond violence.” Good, great. Innovative games that are fun without using violence in their designs are awesome.

Journey screenshot

Journey is one of the seminal games in the recent wave of “artistically-valuable” indie games.



However, this easily morphed into a reactionary movement. Since these games without violence or combat were touted as being somehow better or “more elevated” than your run-of-the-mill murder simulator, it became obvious that a game that was violent was inherently less.

Obviously, this sort of indiscriminate valuing of non-violent games is a terrible idea. A game that doesn’t use violence can be poorly designed and not-fun (Dear Esther, Mountain), just like a game that uses violence and combat can provoke deeper aesthetics (Hotline Miami, This War of Mine). Part of the problem is that nobody has developed the proper critical skills to analyze these non-violent, pacifistic games. Champions of “games are art” too frequently praise the games for not using combat, rather than evaluating the game holistically and praising good design choices. On the other side, core gamers are immediately turned off by the lack of combat and write it off as boring.

This War Of Mine screenshot

Refugees have said This War of Mine accurately conveys the constant fear of living in a war-torn region.



One result of this dysfunction is the proliferation of so-called “walking simulators”. These are games whose main play involves walking around consuming either written, visual, or aural media, perhaps with light puzzle-solving mechanics (or similar accents). Many enterprising developers, whether they realize it consciously or not, have seized on the fact that making such a game guarantees some measure of success. They will be praised by academics and critics interested in furthering games as a legitimate medium, and have their game purchased by the small-but-steady audience of non-core, non-casual gamers (most of whom probably chafe at being called gamers).

Some walking simulators are great; I actually enjoyed Gone Home, in a way that I probably wouldn’t have if it had been a movie. They do a good job of immersing you in a focused, meaningful experience. Others are scattered or diluted by dissonant design decisions — like Corpse of Discovery. But nobody cares, because these games aren’t being evaluated on their merits as a game. They are either praised for being a game without combat mechanics, or they are ignored because they are a game without combat mechanics. Little else tends to go into the evaluation process.

Gone Home screenshot

Gone Home gives the player a meaningful experience despite being limited to looking at rooms and listening to audio.



A student game at USC, Chambara, got changed during development to be “non-violent”. The game originally saw samurai dueling in a starkly colored world. Now instead of blood, hitting an enemy produces a burst of feathers. Apparently this one tweak now qualifies it as “a transcendently beautiful and artistic entertainment game with a pacifistic outlook”. That is a direct quote from a faculty member at the school. You may see why this is troublesome to me. First of all, changing blood to feathers doesn’t change the fact that your game is about sneaking around and hitting other people with sticks before they hit you. That seems a far cry from a “pacifist outlook”. Second, this change actually hurts the game aesthetically. The blood splatters beautifully complemented the dichromatic nature of the game’s world. I consider the stark look of a blood splatter to be more artistic than a burst of feathers. Yet the game’s devs decided to make this tweak. Did they do it because it would benefit the game? No. According to the devs, “we were uncomfortable with the violence the game displayed and did not feel like it accurately reflected who we were and what we believed.” In other words, they value a game that contains bloodshed differently than a game that does not. Are they allowed to make this decision based on their personal beliefs? Absolutely. But isn’t it absurd to pretend that this tweak lends the game a “pacifist outlook”, and that it in turn allows the game to transcend to the angelic ranks of non-violent video games?

Blood Splatters

Blood splatters…


Feather Splatters

…and “feather splatters”.



I would urge critics and academics to judge pacifistic games on their merits as a game, not on their merits as a non-violent game. I would urge developers to treat the presence of combat and violence as just one among a countless sea of other design possibilities. If it aids your experience goal, you should include it and tailor it to the needs of your game as an experience. If it doesn’t don’t include it. But don’t decide to make your game non-violent or exclude combat mechanics just because it means your game will be valued as inherently better by a specific set of people.

Filed under Analysis, Blog, Games Tagged with , , , , , , , ,

← Older posts