Two weeks after the last thread seems like a good time to evaluate how Twitter is going and whether it's valuable.

I don't think Twitter is actually helpful for making me more "fox-like" and less "hedgehog-like", at least when it comes to political & social reasoning. 1/n

There are lots of reasons for this, but I think the most important is that the ratio of [genuinely insightful takes] to [dunking on a strawman of the other side] is really low. This is obviously not unrelated to the character limit. 2/n

But it's also to do with the feedback people get. @SpencrGreenberg is pretty well-known, has a reasonable number of followers, and (imo) has the most consistently insightful tweets of anyone I follow. Very few get much engagement at all. 3/n

This effect has actually meant that in some cases, following people with views outside my real life/other social media bubble has pushed my thinking away from them, despite the intention of my following being precisely the opposite. 4/n

What does seems unique about Twitter though, is interacting with incredibly impressive people, in a way which doesn't really seems possible in any other form. @davidmanheim made this point in my last thread, and I'd include him in the set {incredibly impressive people}. 5/n

But since then, as well as him, @galendruke has responded to a question I asked. I had an interesting conversation with @StefanFSchubert about presenting probabilities. @SpencrGreenberg has responded to me more than once, @PTetlock retweeted a observation I made. 6/n

And I've interacted with a bunch of EAs that I didn't include on the list above because it's plausible I'd have managed to contact them in other ways (probably only 1 degree of separation), but they are still really smart and "I could have" != "I would have" 7/n

This last effect is admittedly slightly difficult to separate from an effect of lockdown though. Physical meetings not being preferred has, I think, had the distinct advantage for me of having had really interesting conversations with people I wouldn't have otherwise. 8/n

Essentially just because I didn't consider video calling an option, so if they lived somewhere else there wa no way I'd have met up with them, and even in London the meeting would need to be 3x as valuable for each of us when you factor in journey time to be worth doing. 9/n

I think this hit home hardest when a few weeks ago I had back to back calls with @fianxu and @jungofthewon, both about things in the forecasting world, and afterwards I just sort of went: Wow, there are so many smart people doing interesting things and some of them... 10/n

... are willing to speak to me about those interesting things! Again it's not specifically twitter, but @LinchZhang is on this list as well, as are several others who aren't on Twitter. Online forecasting co-working is another thing I wouldn't have tried without lockdown. 11/n

Despite some of the above being lockdown related, I do think twitter is uniquely good for this kind of networking, and it's this effect which now dominates my estimate of its value. 12/n

I think I'm going to take a break from Twitter in 2-weeks' time to evaluate, but I'm giving myself another two weeks to get better at a few of things:
- Using #hashtags effectively.
- Finding other things like turning off notifications which improve my user experience. 13/n

- Following more people (please help with this if you have suggestions).
- Unfollowing more people (no need to help with this). Especially given the above, I'm going to try to do this more for the sake of my sanity. 14/n

- Gaining more followers? (I'm not actually sure if this is a valuable goal to aim for, other than that I feel good when people follow me)
- Writing tweets which get more engagement? (see above but moreso, all my most popular tweets have been joke comments) 15/n

- Writing better tweets? (The measurement difficulty here is obvious)
- Making more predictions? (not sure if significant value above just using @metaculus) 16/n

My updated predictions (old prediction in brackets):
Twitter is overall +EV: 60% (60%)
I continue using it: 80% (75%)

Difference in 2nd prediction because I've changed my mind about where I think the EV is coming from, and this could easily happen again. Diff is still worrying.