Fomo3D: A game where everyone is against you

Fomo3D is the name of Dapp that has been running for over a month. As stated on the website, it is a “psychological social experiment in greed”.

In a nutshell, Fomo3D or F3D is a smart contract that has a fund, keys to buy and a timer. Keys cost Ether and add money to the Pot. The timer counts down each second and is increased by each key purchase. When the timer will reach zero, the pot will go to the player who bought the last key.

Currently, there is nearly 22 thousand of Ether in the pot, which is equivalent of $6.5M. The pot has been growing by about 4 ETH per day.

Game Theory

The game resembles War of Attrition, a game theory concept introduced by John Maynard Smith. In this game, two players make small bids continuously over a period of time to compete for the prize, until one of them stops. When this happens, the one who stopped making bids receives nothing and the one who continued receives the prize. However, there are two important features. Unlike a usual auction, the loser doesn’t receive his money back. Moreover, the bidding is scattered in time and looks more like a stream of bids rather than a deterministic choice.

A Dollar auction can be an example of War of Attrition. In a Dollar auction, there is a prize of $100, and bids start at $1. The player who makes the highest bid wins and takes the prize, although the second highest bidder must pay too.

At first, it seems like bidding a small amount definitely leads to a net gain. Other players think the same, and make bids too. Eventually, the highest bid becomes $100. Here, the game reverses. It isn’t about winning anymore, it’s about losing as small as possible. If you’ll win the auction, you’ll lose B – 100 dollars, where B is the highest bid. If you won’t, you’ll lose B dollars. The best option seems to always make a higher bid.

However, that’s not a rational move. That is, if another player follows the same strategy, the bid will perpetually increase, as will the loss of each player.

In other words, it’s most rational to never even buy a single key, no matter how big the prize. Then why have people bought over 34 million keys worth more than 20k ETH? Well, because of irrationality and additional incentives.

First, people make irrational decisions at every turn. Some think they’ll likely win by buying a single key. Others just gamble, like they would do in a casino. Second, F3D is not a textbook example of War of Attrition. You can enter the auction at any moment, meaning that bids are different for each user. The prize is dynamic, as it increases with each bid. Finally, there are additional incentives set in the game.

Incentives

1% of the value of each key goes to the airdrop fund. Initially, the idea of an airdrop was to provide a quick little lottery while accumulating a much bigger and slower pot. In a couple of weeks, Ethereum developer Péter Szilágyi discovered a vulnerability in the smart contract that allowed anyone to write a contract that always wins the airdrop.

Another motive to purchase keys, at least before the hype, is the dividend system. For example, each player of Team Snek (more about teams later) distributes 56% of the key price back to the previous purchasers. The idea of receiving passive income from what can be an endless game sounds tempting.

Finally, there is the psychological aspect. Humans are not rational, so they make irrational choices. The prize is huge, and the cost of the single key is small, so for some players, it turns into a gambling. Those who already bought a handful of keys may experience Sunk cost fallacy, where spending money without any result justifies more spending, even though it will eventually lead to more loss.

Features

Fomo3D has two other features that are worth mentioning. It has teams and the referral system.

Players can join one of the four teams. Teams vary in how Ether distributes between different funds. When a user purchases a key, only part of it goes to the pot. Everything else goes to the affiliated person, developers, dividends, and the airdrop. Exact numbers differ for each team.

By the way, only 48% of the final pot goes to the winner. Everything else is distributed to the developers, next round fund, and dividends.

Another feature of Fomo3D is the referral system. Players can invite others. Each new player that came via player’s unique link will become his referral. 10% of referral’s spending goes to the player.

Inside the contract code, there is a check that verifies that player doesn’t use his own account as a referral. However, there is nothing that can stop someone from creating another account. As there is direct incentive to do so and keep 10% as a kind of cash back on the key purchase, it’s hard to estimate how much the referral system helped to gain initial traction.

Stats

It’s interesting to look at the game statistics and see how F3D grew over time. At the moment of writing, the game was up and running for 40 days.

Current statistics can be seen at F3D website. To get stats for previous days, we’ll use Infura node and getCurrentRoundInfo function of the contract.

Looking at ETH volume growth, the first thing that we notice is enormous growth on July, 20. Why? I couldn’t find any explanation. But looking at statistics, everything — players, transactions, volume — rose at almost the same rate, so I assume it simply went viral.

Another thing that is worth to note that the pot-to-invested rate is nearly constant and equals 22.5%. This makes sense because as we’ll see below, most of the players prefer team Snek, where only 20% of invested ETH goes to the pot.

The number of players correlates neatly with the growth of transaction volume. Additionally, it correlates with the number of transactions made, and the transactions-per-player rate is mostly in the range of 1.5 and 3.

Here’s the distribution of invested ETH by each team. As noted above, most of the money goes to the Snek team. The percentage of Ether invested as Snek team is stable and equals to 88.5%, except for the first few days, where it grew from 59% to 85% in about 10 days. As for other teams, they are almost equally split at about 3-4%.

The distribution of method calls is also quite interesting. For the first 30 days, people invested and withdrew almost equally. Then, there are mostly buys and reinvests.

It’s worth to note that this graph is based on the number of transactions, not the ETH spent/received. It means that 1 withdraw can be equal to 10 or even 100 buys in terms of ETH volume.

Conclusion

The question everyone asks is “Will it ever ends?” Most likely, yes. Unlike closed systems used in game theory, F3D is a real-world application. In the real world, nothing is perfect: neither the code itself, nor platforms, nor human behavior. There might be an exploit in the contract, either intentional or accidental. The miners can collude. Someone can try to DDOS Ethereum while being the last person to buy a key. Finally, people can simply fail to buy a key, which will stop the whole thing (although with an ever-increasing prize this seems unlikely!). There is a whole range of options, from the nonsensical to something more feasible, but it requires only one outcome to stop it entirely. Basically, the shutdown of F3D is a Black Swan event.

Of course, even if nothing of that will work out, it will stop anyway. The reason is that F3D is locking part of the total Ether supply. The price of keys constantly grows. At some point in time, the mining rewards will be smaller than the cost of the keys required to keep the game live. From then, the supply of ETH that is not locked will slowly go down to zero. Thus Fomo3D can’t run for eternity even in theory.

One interesting idea that was proposed is to create some kind of smart contract from which funds can be spent only on the game. If the game will be won by this contract, the funds will be unlocked. One needs to send a few thousands of Ether to this contract and tell everyone else about it. That way, anyone will be able to validate that there is a commitment by someone to spend a lot of money and they will less likely to compete for the prize. This idea might work, but it also requires locking a lot of money which is risky.

Then, someone might ask “When it will end?” and “What will likely be the reason for the shutdown?” That’s harder to answer. Personally, I think that the answer to “when” is this year or maybe even next month.

No matter how long it will be live, it is indeed an interesting experiment.