Why Everyone Should Know Algorithmic Game Theory!
Algorithmic Game Theory (and Mechanism Design) is an exciting branch of economics and computation, that anyone can learn and use. Think of it as designing your own real world game with rules, incentives and socially beneficial outcomes. Here is an overly simple (sort of) introduction to the basic concepts.
As a 10 year old kid, on bored afternoons, I made up a game, a form of catch for my two younger siblings in our living room. Immediately, I was faced with a series of tough conundrums. This may be familiar to many of you!
As the living room was quite small, it ruled out running and catching.
I made up the rules for the chaser, who while touching a piece of furniture should simultaneously establish contact with the runner to make a ‘catch’. This, I felt, was a better game than running and catching. However…
Now, at the time, I was the tallest, fastest of the lot and had the longest reach. I could run around the furniture or reach underneath with ease and speed. Naturally, I was the chaser very few times and got caught even less. As chasing was the least popular option, my siblings were losing interest.
So, I had to make the game fairer by increasing my sibling’s chances of not being caught. So, I made up ‘safe zones’ nearer to the furniture, where a runner could stop and not be caught. The chaser (me!!) would have to immediately move to the next runner.
The problem now was that the runners could opt to stay in the ‘safe zone’ for ever! So, if two runners were in the zone at the same time, the new rule was that the slowest to count to twenty five had to get out of the zone. (I still had an advantage in the safe zone, but my ability to count to twenty five faster more or less compared to my ability to run faster, so this advantage is small!!)
I also had to increase my chances of getting caught to make it fairer. So, my siblings could crawl under the furniture to catch, while I was barred from doing so! (This was a bit tough, because I had to lose a lot of fun privileges!)
At the end of each round, all players got to take a bite of a snack on the dining table — if they individually chose.
Phew!
If you understand this game, you can probably design an algorithmic mechanism in economics and computation, such as, even a blockchain! Here is how it goes.
To design a fair game with a socially desirable outcome(for the game above, this could mean one where the infighting is the least and where they have the most fun points!):
- First, get to know the strengths, weaknesses and preferences of the players (call them types). Either you do the analysis or get the players to report these to you. (They may lie, but as it turns out, that it really doesn’t matter as per the Revelation Principle.)
- Design a game mechanism (rules above such as furniture touch, safe zones, twenty five count rule etc)
- Define payoff for each type (for the game above, it could be fun points for a successful chase or the longest run!).
- Play the game!
As you may know, all games have some form of Nash Equilibrium, i.e. reach some point where they no longer care to change their strategy because it doesn’t benefit them to do so. In the game above, the participants may decide to continue to play as long as they enjoy it or leave because they no longer do so. If the game is designed well, then you may have a fair game with a socially desirable equilibrium- as in above, bunch of kids playing together for most of the afternoon and having lots of fun!
Now to the real, real, world…
As it turns out, mechanism design is one of the most powerful concepts emerging in the recent past, having been used in various areas such as telecom auctions, Google’s Sponsored Ads etc.
I will post another article about how we do the token design of Haggl, which is essentially a mechanism to be used in certain purchase transactions, to maximise the producer and consumer surplus simultaneously for the purpose of total welfare!
Essentially, why I think everyone should know Algorithmic Game Theory is because, it may encourage everyone to think rational, to be fair and design rules for our socio-economic interactions that enable a desirable outcome for all of us! Won’t that be nice!!
— Cheers! (Rajesh Bhaskar — rajesh@trilloc.com)
