Arcade Tokenomics
A practical model for balancing tokenised economies across multiple videogames
My investigation into giving property rights to players has led me to consider videogames from many unusual angles so far. Throughout my investigations however, there has been one perspective I've heard loud and clear, over and over again from many different quarters. It seems that almost everyone believes "it might be possible to make a tokenised economy work within an individual game, but it's quite another matter to make a tokenised economy work across multiple games."
Well, if by "quite another matter" we mean "a lot more complicated" then I'd agree. There are more moving parts, and more moving parts makes the design of any system more complicated; however, if by "quite another matter" we mean "impossible", or anything of that sort then I'd respectfully disagree.
I've seen too many "impossible" features successfully implemented in games I've made to pay much heed to "impossible" these days. In my experience, it usually just indicates a lack of vision to imagine all the ways that it might be possible to achieve what’s required.
Whenever I encounter supposedly impossible obstacles in a game design, I generally reach for one of two tools to find ways over, under, or around them. These are:
1) "Brute force" - does "more" solve it? Doesn't matter what more is needed: more memory, more processor, more people, more time, more anything. If more solves it then it's just a resource problem, so the problem stops being impossible and starts being "how do we secure the resources needed to build it"?
Or:
2) "Cunning plan" - is there another route to our destination nobody's considered before? This is the equivalent of that moment where we find ourselves standing in front of a door we've been pulling harder and harder at, reinforcing our original belief that it's locked. Then our friend/colleague/child arrives and instantly pushes or slides the door open for us, simply because they’re not carrying the burden of our preconceptions and experience. Many figurative doors remain 'locked' for us because of assumptions we make about them, and so it’s important to force ourselves to consider them from different perspectives whenever we encounter them. There are many creative thinking methodologies for doing this, and Edward de Bono’s Six Thinking Hats remains my favourite, especially when I’m working as part of a group.
Based on past experience, I strongly suspect it's this second "cunning plan" category of creative thinking we’ll need to solve the challenges of linking multiple tokenised games together. If it could be solved simply by applying more of what teams have already been doing in the Play-To-Earn and cryptogaming space, then it’s unlikely anyone would consider it “quite another matter”. That being the case, I’m happy to leave the "brute force" approach for others to investigate, and head down the "cunning plan" path for the purposes of this article.
In my previous article I investigated the implications of introducing a tokenisation system to Atari's Pong, to grant players basic property rights in their games in recognition for successful wins.
The system conceived of giving a player one token for winning a game of Pong, and making those tokens the only means of unlocking two new player bats - a Silver and Gold bat. More details of this are available in the article itself, but my most important learning from it was that it showed how game tokens issued with no intrinsic monetary value can develop an independent market valuation purely through their in-game utility. In this way, it confirmed to me that a very simple "real world" economy can emerge spontaneously around any game where the Product Design and Value Design are aligned in such a way as to facilitate this, whether intentionally or accidentally.
In this article I want to extrapolate from a single developer creating a real-world game economy around a single game, to consider how a games publisher with a portfolio of games might be able to create a broader real-world economy across multiple games.
Importantly however, I also want to ensure this creates a set of positive and fair incentives that:
Do not diminish the experience of occasional players
Enable dedicated players to participate fully in a single game of their choice without being forced to pay more than their standard "entry fee", or play a different game they wouldn't choose to
Reward all players with some form of property right benefit they wouldn't get from a standard Free-To-Play game
Create a genuine, sustainable, "real-world" economy around the portfolio of games that delivers additional value for both players and creators through the purchase and sale of tokens
I know that sounds like a very large circle to be squared, but fitting square pegs in round holes has become something of a specialty of mine. It’s a side-effect of spending so many years coaxing performances out of computers that were never designed to do such things. So let's start where I always start - by grasping the first concrete problem right in front of me and solving that to see where it gets us.
The four critical variables I identified in the process of tokenising Pong were:
The value of a token
The total number of tokens
The conditions of token issuance, including rate of issuance, and
The conditions of token redemption, including the rate of redemption
I’m going to assume that since I was able to show how tokenising Pong could cause a real-money economy to emerge spontaneously around it (albeit in a very simplified way), that no one still needs to be convinced that it would also be possible to do this for other classic arcade games.
I’m also going to assume that everyone’s still comfortable with the logic I used to determine a reasonable “Value Per Token” for Pong, and that this could be applied to other classic arcade games without much controversy.
I didn’t feel the need to explore the implications of capping the total number of tokens issued in my previous article, as it didn’t seem relevant to getting the economy functioning. I certainly acknowledge that limiting tokens would have an impact on the economy by skewing incentives, but I also believe these would merely create a variety of interesting flavours, rather than impacting the basic taste of the experience created. In other words, limiting the number of tokens might make it feel more or less fun for players, but it’s very unlikely to make the difference between it working or not working. So again, I’m prepared to leave that aspect aside for now and address it in due course if it should become more relevant.
The conditions of redemption and rate of redemption were also investigated for Pong, which led to the creation of additional Bats for players to unlock. Again it doesn’t seem unreasonable to assume that this basic principle could be extended within Pong if additional redemption sources were required or desirable. The same approach also seems general enough to be applied to any other classic arcade game too.
That being the case, then the first significant barrier to adding more games to the economy would be the conditions of token issuance across the different games. I deliberately chose the simplest issuance condition I could imagine working effectively when analysing Pong, which was to issue a token in exchange for every game won. That worked well enough when considering tokenisation within Pong on its own; however, it's clearly far too blunt a system to be implemented across multiple games without additional refinement.
Another one of the lessons established in my previous article was the importance of building a tokenisation model around finished game designs that had already been market tested to confirm key commercial metrics. Specifically, the profitability of the game as measured by the Lifetime Value Per Customer, or Profit Per Play, etc., was shown to be crucial for establishing a maximum value of each token at issuance. As I currently see it, designing a tokenisation model around a game whose design is still to be finalised and validated by the market, multiplies the commercial risk by at least an order of magnitude. More likely two.
This means that before attempting to expand the tokenisation model beyond a single game it would be useful to confirm the specific games the tokenisation model is to be applied to. For the purposes of this thought experiment, I'm going to complicate the model by shifting perspective from an individual videogame developer with an individual game such as Pong, to that of a small videogame arcade owner with a portfolio of three games, each supplied by a separate business that has developed its game without considering cross compatibility with the others.
To ensure the model can accommodate the additional complexities of games from entirely different genres, the two extra games I’ve chosen will be Space Invaders, and Pac-Man. In this way the game designs of the three games will remain familiar to most readers, but also be as varied as possible. That should stretch the tokenisation model thoroughly enough to ensure it can meet all the functional criteria listed above.
With the three games defined, it becomes possible to solve the tokenisation problem through the lens of finalised and testable game designs. Otherwise it would be necessary to guess the implications of tokenisation decisions against the backdrop of designs that are evolving simultaneously. I know from first-hand experience just how hard it is to design something as simple as audio for a game whose design is constantly changing, so I think I’m also reasonably qualified to offer an opinion on how complicated it might be to design a whole economy across multiple games under those circumstances. Though I am also very aware of the Dunning-Kruger effect, so would be happy to have the debate with someone more qualified.
As mentioned, the token issuance chosen for Pong in the previous article was linked to game completion, in that a token was only awarded for winning a game. This is simple to do with a game such as Pong since it has a well-defined win condition - the first player to reach 15 points wins. Unfortunately, neither Space Invaders or Pac-Man have any sort of win state defined as part of their game design.
Pac-Man has a well-documented end state that happens when the player completes 256 screens, and so this creates a technically defined "win state", as well as limiting the maximum score achievable to 3,333,360 points, but this is not by design.
By contrast, Space Invaders has no known comparable technical limitations, and is currently limited only by the player's skill. According to the Internet’s ‘official’ arcade world record database Twin Galaxies, the current Space Invaders world record of 218,870 is held by John Tannahill and was set on 12/12/2017.
Aligning issuance criteria therefore becomes the first practical complexity when trying to design any sort of token system that would gather and distribute value to players over multiple games. So how might this best be achieved?
The first variable to consider is whether we want to complicate the discussion at this point by assuming that a tokenised version of Pong had already been in operation and was distributing tokens to players on the basis of 1 token = 1 win before the need to create a tokenised version of Space Invaders and Pac-Man arose.
The choice of a token issuance model based on triggering a win state would be an added complexity at that point, given that the designs for Space Invaders and Pac-Man do not contain any sort of comparable state. It would therefore be necessary to accommodate Pong's existing token system within the expanded token system.
I believe this further reinforces the importance of working with as few variables as possible from the outset, as well as highlighting a very real practical problem that game tokenomics designers would be wise to consider upfront as well: namely, what the implications of swapping the tokenomics model from a game for a new tokenomics model would be, and how they could be accommodated in a way that was fair for any existing token stakeholders.
This is a real practical concern that will definitely need to be addressed by Play-To-Earn businesses in future, but I feel it is beyond the scope of the current thought experiment, and best left for a later article once the foundational elements are better understood. For that reason I'll begin with the core assumption that in this particular example we have the luxury of starting the arcade tokenomics design without a pre-existing token system already in the market.
PRACTICAL ARCADE TOKENOMICS
So it's clear that issuing tokens based simply on achieving a win state is unsuitable for use within a more generalised arcade tokenomics model, because only a minority of arcade games include a win state, with Pong being the most obvious example. By contrast, most other arcade games dispense with win states altogether, and only contain well-defined lose states, such as Night Driver's time limits, or Lunar Lander's fuel. This is desirable for game development companies since it enables games to be played by individuals on their own, without the need to recruit a willing competitor before being able to play.
This then poses the question of what issuance source would be best suited to a more generalised arcade tokenomics model, and it turns out there are many possibilities to consider. For example, although it has now become clear to me that win states are not a useful issuance mechanism for arcade games, in that most games do not share this feature, it is not immediately clear what should replace it. Lose states would be one possibility since they do seem to be common to all arcade games, even Pong, but while this might work from a Rules Design perspective, issuing tokens for losses would seem unlikely to stand up to scrutiny from a Product Design or Experience Design perspective. Most importantly, losing is usually what players are trying to avoid doing, and issuing tokens for such events could end up incentivising behaviours that are at odds with each game's core design.
It would therefore be better to look for features shared between as many games as possible, and certainly as far as arcade games go the next most obvious feature would be score. All arcade games have some sort of scoring mechanism included as part of their game design, and this already mirrors the intended incentive structure of each game. Therefore issuing tokens in recognition of score would appear to provide a more reliable and consistent token issuance mechanism across the widest number of games.
By coincidence, it could be argued that this is what my original tokenisation system for Pong was already doing anyway. A token was being issued for each win, but that is exactly the same as a token being issued for every 15 points scored in a game. The issuance is indistinguishable on a practical level; however, the need for a more considered design approach becomes clear as soon as the concept is extended to invite comparisons across multiple games. This demonstrates the value of considering the design aspects of the tokenisation system thoroughly from first principles at every stage.
Assuming the initial Pong tokenomics model suggested in my previous article remains a reasonable benchmark, then the next priority becomes understanding scoring in all the other games to be added to the system. Anyone with even a cursory knowledge of Pong, Space Invaders, and Pac-Man will appreciate that each game awards score to players very differently, which could lead to wildly different issuance across games without some sort of stabilising mechanism to ensure consistency of token issuance. So, at this stage it becomes important to dig into the nitty-gritty details of each game's scoring system.
Pong is by far the easiest to understand. The game finishes when a score of 15 points is reached, meaning the maximum number of points achievable by a player per game is 15.
Space Invaders is more complex. The game begins with five rows of 11 invaders. The 2 rows closest to the player score 10 points per invader, the next 2 rows score 20 points per invader, and the row furthest from the player at the top of the screen scores 30 points for each invader. That makes each screen worth a total of 990 points. From this it is immediately self-evident that issuing a token for every 15 points would cause huge problems for Pong if the same tokens were accepted in both games. That would make it possible for Space Invader players to earn 66 tokens per screen of invaders, requiring a player to complete only 2 screens of invaders to earn enough to be awarded a Silver Bat in Pong, and a mere 8 screens of invaders to earn enough for Pong's Golden Bat.
The problem becomes even worse when Pac-Man is added to the mix. It awards 10 points per dot, of which there are 240 dots per maze. Energizers score 50 points, and there are 4 of those per maze, so by completing one Pac-Man maze a player would receive a minimum of 2,600 points, generating 173 tokens each time if the same 15:1 issuance rule from Pong was preserved.
Of course, unlike Pong and Space Invaders, Pac-Man gives players the opportunity to score bonus points by eating ghosts and fruit. If the player eats all four ghosts another 3,000 points can be gained, and with four Energizers that means an additional 12,000 points can be earned on each of the first 17 levels, after which ghosts stop turning blue and cannot be eaten.
Additionally, fruit appears after 70 dots have been eaten and after 170 dots are eaten, with each fruit being worth anywhere from 100 to 5,000 points, depending on the level. For anyone interested in the details they're available here. So even a Pac-Man player that doesn't manage to eat any ghosts or fruit would generate enough tokens for Pong's Silver Bat from completing a single Pac-Man maze, and have enough for Pong's Golden Bat after completing just three mazes. Any good player, able to maximise their score by eating all fruit and ghosts could generate 14,800 points from the first maze alone, or the equivalent of 986 tokens.
Clearly, at this point, any independent value established through Pong's tokenomics system would be rendered worthless, since a halfway decent Pac-Man player would be able to generate enough tokens to acquire Pong's Golden Bat by completing a single screen of Pac-Man, rather than the 500 full games it would take to acquire this purely from playing Pong.
It would seem some sort of gearing system is needed to align the effort required to earn tokens from each game, and in traditional economies these gearing systems tend to be referred to as "Exchange Rates".
When it comes to establishing exchange rates, I like to turn to the empirical approach used by Isaac Newton (he of "Laws Of Motion" fame) for his work on the gold/silver ratio during his time as Warden and then Master of the Royal Mint.
His approach towards standardising the exchange rate between gold and silver was to find a fundamental relationship between the elements in nature. Doing that would let him define a stable foundation upon which an exchange rate for the two metals could be set, that everyone could verify independently without the need to trust the Royal Mint.
He achieved this through field research and laboratory experiments, eventually determining the naturally occurring ratio of silver to gold within the Earth's crust to be 15.5 to 1. It varies from region-to-region of course, and more modern calculations suggest it's actually closer to 19 to 1, but regional variations aside it demonstrates an excellent empirical way to go about establishing an exchange rate. How then, might one bring a similar approach to standardising exchange rates between games?
As described above, the main issue is the rate at which points, and by extension tokens, are awarded to players in Pong, Space Invaders, and Pac-Man. This is complicated still further by the fact that while the number of points that can be awarded in Pong is limited to 15 it is essentially unlimited in Space Invaders and Pac-Man. 25 cents spent in Pong generates a maximum of 1 token, whereas 25 cents spent in Space Invaders or Pac-Man generates a potentially limitless number of tokens, depending on the skill of the individual player.
The website howlongtobeat.com/game?id=7201 lists the median length of time for a Pong game as 7 minutes, so approximately 2 points per minute, and one token issued every 7 minutes of play on average.
As mentioned above, the current Space Invaders world record of 218,870 is held by John Tannahill and was set on 12/12/2017. It took 3hr 47mins and 44sec to achieve, approximately 960 points per minute, or 64 tokens per minute
Using the empirical exchange rate approach favoured by Isaac Newton, that implies an exchange rate of Space Invaders to Pong tokens of 480 to 1 (960 / 2 = 480)
Twin Galaxies also lists the current world record for fastest "completion" of a 3,333,360 Perfect Score in Pac-Man as 3hrs 28mins 49secs, and was set on 22nd May 2013 by David W Race. That equates to approximately 15,963 points per minute, or 1,064 tokens per minute
This implies an exchange rate of Pac-Man to Pong tokens of 7,981 to 1 (15,963 / 2 = 7,981), and a Pac-Man to Space Invaders exchange rate of 33 to 1 (15,963 / 480 = 33)
Just as with determining the ratio of gold to silver in the Earth's crust however, this is not as exact a science as we would ideally like. That's because the rate of points being awarded varies from player to player, as can be seen by comparing the top 6 "Fastest Completion" Pac-Man scores on Twin Galaxies. They range from David's world record of 3hrs 28mins 49secs to Donald Hayes 5hrs 24mins 46secs, implying a range of 15,963 points per minute down to 10,264 points per minute, and it's safe to assume that if they are the world class scores then average players will generate even fewer points per minute.
On this basis it is clear that more detailed metrics than the desk-based research carried out here would be required in order to have confidence in the exchange rates set; however, while some monitoring and adjustment of exchange rates would be required, the overall approach seems robust enough to proceed with for the moment. It shows that balancing multiple tokenised game economies would be possible for arcade games by using a “tokens per unit time” metric – or, to borrow bitcoin’s parlance, “Proof Of Work”. This seems to form a reasonable initial mechanism by which to establish cross-compatibility between tokenised economies within arcade videogames.
Are there others that could be used? Certainly. Would they be more successful in establishing a stable and fair exchange rate between games? Possibly. Are there likely to be significant flaws in this methodology of exchanging value between games? Perhaps, but regardless, the system described above appears to be a reasonable, practical approach to solving some very specific problems that would otherwise prevent multiple tokenised economies from being combined. It provides one clear example mechanism for achieving it based on game fundamentals, and in doing so provides some confidence that even if it didn't work in practice for any reason, the basic approach remains valid and would be likely to lead to other practical solutions that could be implemented.
It suggests that, just as with Free-To-Play, there will be some unique metrics Play-To-Earn games need to measure, monitor, and utilise well in order to combine economies across multiple games.
Is that desirable? Well, that will ultimately be for the market to decide, but at least it seems possible if it is.
In any case, from what I can tell so far, whoever can build the most trusted system of equilibrating value between tokenised game economies is likely to become a critical part of the infrastructure of Play-To-Earn gaming if the sector becomes mainstream at any point.