Demand for distraction is highly inelastic, for me
A couple of months back I had written about the economics of distraction. Recent observations suggest I'd massively overestimated the elasticity of demand for distraction
In March, I logged off Twitter, and loudly wondered if demand for distraction was elastic. I had drawn some demand-supply curves (rather, got ChatGPT to draw them for me), which showed that getting off twitter (and cutting other sources of distraction) should reduce the overall quantity of distraction.
And then I had written:
Or is it simply that the demand curve for stimulus is far less elastic than what I’ve drawn here? I’m not able to get ChatGPT to draw an alternate, but if the demand curve for distraction were to be near-vertical, then the cutting of sources of supply (such as Twitter or medication) will not actually change the amount I’m actually distracted. All it will do is to make the distraction more costly!! I’ll spend more of my brain power for the same distraction.
Two months down the line, I believe that the demand curve for distraction, at least for me, is highly inelastic. In other words, cutting out a source of distraction does absolutely nothing for my overall distraction levels - I simply find a new source of distraction.
I had to play around a bit with ChatGPT-4o to get the curve better, but finally it gave me this. It’s not perfect but I’ll use it.
The demand (blue) curve here is the demand to get distracted. Notice that this curve looks very different from the one I drew the last time I wrote about this topic.
Green is the initial supply curve - there more I’m “willing to pay” for the distraction, the more the distraction that is available. These distractions could include social media, games, sports, TV and the likes.
When I cut out one source of distraction (such as getting off twitter), the curve for the supply of distraction shifts left (all other distraction costs the same as it did, but one source is eliminated, so for a given effort / cost, the distraction available is lower).
My initial hypothesis (when I wrote that post in March) was assuming that my demand for distraction was sort of elastic, which meant that cutting a source of distraction (going off social media, for example) would result in cutting the overall amount of distraction.
However, a lot of recent experience is suggesting that my demand for distraction is far less elastic.
Every few days, either my wife or I recognise that I’m getting obsessed with something, and then try to cut out that source of distraction from my life. Twitter is one thing. Watching quizzes on YouTube is another. Playing online chess is yet another. And so on.
However, of late, what I’ve noticed is that every time I cut out one of these sources that I was obsessed with, I’m able to get over this source (I haven’t played online chess for some 7 years now), but I start going elsewhere seeking distractions. When I’m off Twitter, I find myself checking Instagram and LinkedIn (!!) far more often. When I’m off online chess, I spend more time reading the bridge column in the Times Of India. When I’m off YouTube, I spend more time on Twitter.
Whenever I cut out one source of distraction, I go around seeking another
In other words, cutting out a source of distraction is only a temporary fix - my demand for it is such that I’ll go elsewhere looking for it. This suggests to me that maybe I should simply not bother blocking out any sources of distraction? That way I’ll be able to get the distraction for cheaper (look at the first curve above)!
Tailpiece
This is a highly ill-formed thought so I’ll just leave it here as a tailpiece. Maybe ChatGPT displays symptoms that one might associate with ADHD. In order to generate graphs for this newsletter, I was trying out several combinations and prompts. In one, the explanation it gave was absolutely perfect, but the graph was rubbish.
It almost meant that by the time it went from explaining the solution in text to producing the graph (remember that LLMs predict one token at a time), it had forgotten what it had said in the text!
Ages back, I had experimented with some mild substance and found that it made my thought much more disconnected. It was like ADHD taken to yet another level. Now, LLMs have context - the prediction of the next word token is dependent on all the words tokens it has seen / generated so far. However, beyond a point maybe some of the earlier context gets “lost” and it starts saying contradictory things.
Now I’m wondering if LLMs might offer some insights into research on ADHD.
This reminds me a lot of how anxiety works for my own brain. It really feels like if I can just solve whatever the THING is that I'm worrying about, then I'll stop worrying forever and everything will be fine. But then I solve the thing...and my anxiety just finds some brand new shiny thing to worry about! Obviously external stuff does matter, and reducing major stressors definitely helps out, but also there is an internal component where I have to work on addressing the anxiety itself, not whatever the current external stimulus is.
You have messed up the Econ 101 here. The commodity that is being priced here is distraction as a whole, not distraction via a specific channel, Twitter or whatever. The price you are paying is the effort you are putting in to obtain the distraction. As you yourself mention, when you gave up Twitter, you were able to easily obtain Reddit. This means that you did not in fact have to pay extra to obtain a supply of the commodity. Your experiment does not prove that the demand for distraction is highly inelastic. To test the hypothesis properly, you need to find a way to reduce the supply of distraction. Put away your phone and maybe even your laptop and try to write on paper. Then you'll find out how inelastic the demand is.