Twitter is a slot machine composed of humans

Engineers design slot machines to be addictive. They exploit the phenomenon, first identified by B.F. Skinner, known as the “variable schedule of rewards.” It turns out that behavior is most effectively reinforced not when it is consistently rewarded but when there is a random element. When you pull the lever of a slot machine, you might win but you might not. Engineers figure out just how many times you need to win to keep you going, and build that into the algorithm. The algorithm will give you a large reward (the jackpot) in order to keep you going. But small rewards at more frequent intervals are necessary because too many jackpots would degrade their value (as well as bankrupt the casino). “Near misses” (the jackpot symbol is visible but not on the “payline”) can be constructed on the slot machine through algorithmic manipulation—they fool people though we all know abstractly that from the algorithm’s perspective a near win is no different from any other loss. But they are psychologically effective; that is why they appear more often than random.

Twitter works on the same principle. When you send out a tweet, the reward is variable. Depending on the number of followers you have, your tweet will normally not be retweeted or retweeted only occasionally. From time to time, you will hit the jackpot—a media personality (or his or her factotum) retweets it to thousands of followers, generating a cascade of retweets and likes. At other times, you receive a small reward of a few or dozens of retweets or likes. I suspect the power curve strongly resembles the slot-machine version.

Twitter builds addictive loyalty in other ways. The more followers you have, the less you want to lose them. As you scroll down through the increasingly boring tweets, you’re informed that new ones await you at the top of the screen. And then the tweets you receive also follow the variable reward system. Most are boring, some are interesting, a few are great.

Slot machines are entirely self-contained; that is why they are so effective at ruining people. Twitter is a decentralized form of slot machine, one that uses nominally autonomous human beings as its vectors. That is also why Twitter is not as addictive as slot machines. People aren’t as reliable as algorithms are. But I suspect Twitter’s engineers have caught on to this problem. Twitter itself tries to encourage people to write good tweets rather than bad tweets. I have often wondered whether Twitter creates artificial followers for people, or randomly gives prominence to their tweets. If not, it should (that is, from the standpoint of maximizing addiction). Twitter engineers should figure out a way to construct a “near miss” as well.

I sense that people are catching on that social media is a con. Maybe not all of it. Instagram—definitely yes. There, the addictive quality is reduced to its purest essence, with people simply posting to get likes, and liking to get more likes, and everyone caught in a Red Queen-style competition to nowhere. Twitter has much to learn from Instagram. Facebook, it’s not so clear. But the important point is this: a social medium does best by addicting its users, and so there are immense incentives for it to do that. Backed by vast sums of money, limitless data, and huge engineering talent, sooner or later these companies will figure it out. If you are not yet addicted, just wait for the next update.

Some people have said to me that they find value in Twitter. The major claim is that they receive useful information from reliable sources very quickly. Of course, if every element of Twitter were literally valueless, it would not have gotten off the ground. A slot machine, too, pays out once in a while. But unless you are a journalist with a breaking-news beat, ask yourself how important is it that you get information two minutes after an event rather than a day. Many other services exist to supply people with specialized knowledge that they need for their work. Bloomberg BNA’s Law Week emails tell me what I need to know about the latest law news. They couldn’t be less addictive.

I suspect the real reason that the chattering class—the academics and journalists who filled my feed—uses Twitter is that they are afraid of falling behind. If everyone tweets but me, am I no longer important? Does my opinion no longer matter? In this sense, Twitter is the Instagram for intellectuals—a negative-sum game that users play purely because everyone else does.