Newsletter
Bias vs Noise pt. I: Bias vs Bias and other things
March 21, 2025
Hello,
Here’s everything since my last little missive to you:
Bias vs Noise pt. I: Bias vs Bias
Excerpt: The perils of cognitive bias is a subject that’s dominated a substantial slice of social psychology, and appears in any leadership or personal development course as something to be avoided at all costs. It’s interesting, but it’s not actually that useful. You can’t sift through 200+ biases to work out what you might do wrong. The brain treats bias differently. Bias is a strategy to solve certain kinds of problems. Let me show you how.
Main idea: The behavioural economists treat bias as an error. But the brain isn’t an economist. It’s more like a statistician, using bias as a trade-off. Bias ignores noise to see something more clearly, though of course, sometimes the noise shouldn’t be ignored.
Speculative claims that humans might have evolved from two ancient ancestral populations that:
first diverged from each other around 1.5 million years ago and later merged back together 300,000 years ago, initiating a genetic mixing event that culminated with the birth of modern humans.
Here’s a more accessible journalism piece on the paper.
–
Refreshing article about the obvious faction of the right, and the more subtle one:
a stark division is revealed between those who opposed wokeness primarily because it was authoritarian and being institutionalised and those who opposed it because it was the wrong kind of authoritarian and wanted to institutionalise something else
She goes on in her follow up article to describe this illiberal right-wing counterpart to the more frequently criticised ‘woke’ left in more detail.
Worth a skim. Or, she also did an interview.
–
It’s not quite cognitive dissonance, but AI can rationalise stuff to itself too:
Chain-of-Thought (CoT) reasoning … is not always faithful, i.e. CoT reasoning does not always reflect how models arrive at conclusions … on realistic prompts with no artificial bias … Specifically, we find that models rationalize their implicit biases in answers to binary questions (“implicit post-hoc rationalization”). For example, when separately presented with the questions “Is X bigger than Y?” and “Is Y bigger than X?”, models sometimes produce superficially coherent arguments to justify answering Yes to both questions or No to both questions, despite such responses being logically contradictory. We also investigate restoration errors (Dziri et al., 2023), where models make and then silently correct errors in their reasoning, and unfaithful shortcuts, where models use clearly illogical reasoning to simplify solving problems in Putnam questions (a hard benchmark).
–
How business metrics broke universities. I hate the layout of this article, but suffer through. Interesting skim:
Small departments were eliminated or merged into new units that employed dozens of adjunct instructors teaching hundreds of sections across multiple locations. Demands for efficiency and scale have led to the replacement of senior faculty mentorship with online training modules. Regular departmental discussions, collaborative curriculum development, shared teaching experiences—all of which had a politically moderating function—disappeared.
When the decision to mount a suite of courses is driven by metrics, the rigor of each class matters less than its ability to attract students. Radical voices that spark controversy suddenly have an advantage. Assessment coordinators can point to high enrollment numbers and enthusiastic student feedback as evidence of success. Quality and rigor do not matter. And when departments are dissolved or merged, the traditional role of senior faculty in mentoring junior colleagues has been replaced by centralized “course development” training programs, and their influence over hiring and promotion is diminished by administrative mandates.
And most interestingly to me, they propose it as an input to the increasing politicisation on-campus:
It was not obvious at the outset that centralization and bureaucratization would drive politicization, but perhaps it should have been. With departmental homes broken and disciplinary ties severed, why wouldn’t faculty seek emotional connection in politics and causes? Why wouldn’t they spend their extra time on social media rather than in the lab or the library?
Though, obviously the solution to this wouldn’t be to go back to the old problem they were trying to fix, so I feel like this article is missing something.
–
We are running out of eggs. It’s bird flu. I was a bit worried about the evidence it might spread to us, but now I’m mostly worried about my eggs.
–
The Managerial Class Has No Future:
The professional-managerial class (PMC) comprises highly educated professionals who work in fields like law, medicine, finance, and corporate management. Unlike traditional elites who can pass down family businesses, land, or powerful social networks, the PMC’s only transferable asset is often their earning potential–which must be painstakingly re-earned by their children through similarly grueling educational and professional hurdles. …
There is an absurdity to the results – parents queueing up for private kindergartens so their children may fingerpaint in prestigious company, high school kids earnestly talking up the fashionable nonprofit they founded with parental funding, eight year olds diligently practicing dressage so that they may study at the best institutions. But zoom out and the picture is a grim one. The professional-managerial class has no reliable means of reproducing itself without being taxed by these institutions, and however high the toll, they must pay. …
The result is a class that, despite its high incomes and social prestige, is fundamentally unsustainable. Without the ability to directly pass down their status, and with institutions continuously raising the price of admission, the professional-managerial class finds itself locked in a cycle where each generation must start from scratch – until, inevitably, the system consumes more than the class can produce.
But the result isn’t that they get priced out and a meritocracy emerges. It’s that everyone gets priced out. They have ideas about what to do, but obviously people will do nothing. Interesting thought experiment to think about what would happen by default.
–
A Brief History Of Accelerationism. What started as a neo-Marxist vision of breaking capitalism by moving through it:
what the accelerationists affirm is the capitalist power of dissolution and fragmentation, which must always be taken one step further to break the fetters of capital itself
Is now the label we apply to techno-optimists. Are they winning then? Interesting cocktail trivia, anyway.
–
I hope you found something interesting.
You can find links to all my previous missives here.
Warm regards,