shakoist Profile picture
Apr 28 19 tweets 4 min read
I want to talk abut my framework for how to evaluate thinkers, since most people are not very good at reasoning through how to do this.

people evaluate ideas from a person independently, rather than as a portfolio. and they get super emotional and personal about it.
what if we treat the ideas from intellectuals not only on their own merit, but based on how they contribute to a portfolio of all idea generating thinkers?

e.g. we know Apple is a 'really good' stock, but in a portfolio of assets you benefit from a mixture of uncorrelated assets
similarly, it's possible to pay close attention to some intellectuals, even if they are mostly wrong and stupid, if when they are correct, they're correct in a way that's uncorrelated with most other people you follow.

let's start with Ann Coulter.
Ann Coulter needs no introduction. She's interesting though due to her rare form of girl-autism, where she honestly doesn't care if her fans like her. She was one of the first pro-Trump supporters (she wrote a book on how much she liked him) to subsequently reject him.
That's interesting. Why would someone who was such a fan reject him? And with Coulter, you know she has no real likeability bias.

So in the cluster of right-leaning thinkers, she can be a powerful leading indicator. Even if she's 'wrong' more than others.
Okay how about @nntaleb . A lot of people make the mistake of thinking they need to "like" or "dislike" Taleb. Actually you don't need to like or dislike him. Instead treat him as an asset that generates ideas.

His object level takes are often controversial.
but unlike most thinkers, this guy has spent decades enumerating his entire set of primitives, heuristics, and probablistic heuristics he uses to figure out problems.
I personally think Monsanto engineering food to have more nutrients is a good thing. But Taleb can point to a lot of work he's semi-formalized on precautionary tail risk to justify his beliefs.
read his primitives and his object level takes, and incorporate them into your meta-model appropriately.

He's always reasoning from first probabilistic principles. who else in your portfolio does that? Probably no one. he deserves a large weight for this reason.
We also have @Noahpinion . A lot of people dislike him, but that's because you don't know how to read him.

first of all, the dude is prolific at consuming vast amounts of information and analyzing it through an economic lens.
but about 10% of his stuff is definitely a 'reason slave passions' type deal, where he rationalizes arguments in support of his progressive beliefs.

so in your meta-model, downweight anything he writes about that is progressive coded, but don't discard the rest.
--that's the beauty. just treat him as a piece of software with a certain bias. don't take it personally. don't get upset. just incorporate him where appropriate, and disregard otherwise.

(I subscribe to his substack, his writing speed is impressive)
who else is interesting...

oh, here is how I read @tylercowen -- my meta-model is that 1. tyler regularizes a lot of his beliefs pretty intensely based on his perception of uncertainty. and 2. on controversial topics he is Straussian.
so if tyler writes something with a clear object level belief, I weight it much higher in credibility than nearly anyone else.

and if tyler hints at something subtly AND it's a controversial topic, I will allow myself to 'read into it' more than I would for other thinkers.
(tyler is also relatively good at not overly obsessing over the current cultural thing. he operates more like an MA(12) so if he's not writing about it, it may indicate it's less important than you think)
@mattyglesias for some reason rubs people the wrong way. Some people think he's based, and then get upset when he unveils some progressive trait they hate. or get upset because he entertains ideas that are too right-wing.
but Matty is pretty good at sober left-leaning analysis that is long-term oriented, not based on short-term opinion.

So in your model of the world, his takes can be trusted to not be as influenced by wanting to be liked by other left-leaning influencers.
The cool thing about this approach is you benefit from a wisdom of crowd/mixture of experts type deal.

It's better to have a portfolio of thinkers, each of whom is kinda weird and autistic (the good kind), and take a weighted average of them, than to trust 'experts'
even IF a single expert would dominate a single draw from your portfolio of autistic thinkers. But due to wisdom of crowds, your custom portfolio will beat standard so-called experts.

• • •

Missing some Tweet in this thread? You can try to force a refresh

Keep Current with shakoist

shakoist Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!


Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @cauchyfriend

May 14
I always wanted to write about how I learned2code, so this is good prompt to explain my process.

opening scene: age 21, graduated with a degree in econ and political science, doing excel monkey work as a permanent-intern at a 2nd tier asset mgmt firm.
I had never written a line of code at this point. I'm generally not thrilled with how my first year out of college is going. On the heels of the great recession, I have a middling job, and it's not even interesting. I realized coding is the tool
that underpins everything I want to do (deep analysis, modeling, stats, econometrics), but again I have zero experience with any actual modeling or coding. I hadn't even taken a full undergrad math sequence or intro CS or anything.
Read 32 tweets
May 13
Bootstrapping up new systems, whether financial, AI, or policy based, then immediately doing population wide alpha testing, is actually incredibly based, and it's how you learn how to rapidly iterate and improve, despite being painful.
Even though it seems like all people know about Popper these days is some lame version of his "paradox of tolerance", his political philosophy was that it's basically too hard to guess what is going to work, so we should just try a ton of small experiments.
A lot of these new systems or transformative tech just can't exist without being experimented with en mass. Tesla can't work without insane amounts of training data. Ethereum can't work without being battle-tested by millions of people.
Read 5 tweets
May 10
so here is sort of a ridiculous way I think of market equilibrium. despite having worked in/done financial asset-pricing research, I keep going back to this cyclical model that I call the Gargoyles of Capitalism.
we all know Adam Smith was gesturing at something with the invisible hand. there was some equilibrium process emergent among humans that guided individuals to make certain choices. almost as though it were orchestrated, yet completely decentralized.
the gargoyles of capitalism operate differently. They grow disgusted and angry when too many people grow fat and lazy.

if a grifter here and there gets rich, and lives a life of unproductive luxury, the gargoyles may not notice. but when too many begin, their rage boils.
Read 13 tweets
Apr 10
A few predictions:

The fed will get inflation back under control, and the 5yr5yr won't go above 3.5% [probability 85%]

rationale: fed is still staffed with serious people. even if their predictions were bad, they will adjust course.
The murder rate will be in a new regime, and won't go back to 2015-2019 levels over the next 5 years. [70%]

We are in a new equilibrium of violence. The previous equilibrium involved less law enforcement, because there was a credible belief you would get caught for crime
Now that we are in a new equilibrium, the only way to go back would be a huge surge of law enforcement, and state sanctioned violence. The populace won't support this, so we are stuck in a bad equilibrium.
Read 5 tweets
Apr 8
Looking to see if I owe taxes on my parental leave.

Washington State parental leave page says "Sorry, we don't know if the IRS considers this taxable, and the IRS won't answer us"

lmao Image
This isn't a goof Image
Read 4 tweets
Feb 6
I've been deeply concerned observing so many of you categorize verbal rotators as wordcels. Or worse, claim that a high SAT math/verbal is isomorphic to rotator/wordcel. My friends, this is *wrong*.

I've attempted to clear it up for you:…
@tszzl says he doesn't fully agree with Balaji in this tweet.

If wordcel simply means high verbal IQ compared to math, it's an incredibly boring taxonomy. We don't need to recreate SAT scores.
To me, it's an epistemic question. A rotator is trying to make sense of the structure of reality. It doesn't *have* to be with code. You can do it with words as well.

I make a case that Tolstoy is a rotator. War and Peace is essentially an agent-based simulation.
Read 5 tweets

Did Thread Reader help you today?

Support us! We are indie developers!

This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!


0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy


3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us on Twitter!