HNHacker News
TopNewBestAskShowJobs

RA_Fisher

891 karma · joined September 14, 2012

I’m a statistician-economist and I was Zapier’s first data scientist & 9th employee.
submissionscomments
RA_Fisher··on Cowork: Claude Code for the rest of your work
Depends if the job requires a lot of information and the person is excellent at what they do, bc then AI augments the worker more than substitutes them.

But for many people, yes, AI will mostly substitute their labor (and take their job, produce operating margin for the company).

RA_Fisher··on LMArena is a cancer on AI
Yes, for sure! I can think of a few ways.
RA_Fisher··on LMArena is a cancer on AI
100% (am a Bayesian statistician).

Isn’t it fascinating how it comes down to quality of judgement (and the descriptions thereof)?

We need an LMArena rated by experts.

RA_Fisher··on AWS raises GPU prices 15% on a Saturday, hopes you weren't paying attention
Demand for AI is growing. It makes sense they’re trying to increase profits.
RA_Fisher··on Why does a least squares fit appear to have a bias when applied to simple data?
The issue in that case is that OLS is BLUE, the best linear unbiased estimator (best in the sense of minimum variance). This property is what makes OLS exceptional.
RA_Fisher··on AI will make formal verification go mainstream
LLMs are okay at bisecting programs and identifying bugs in my experience. Sometimes they require guidance but often enough I can describe the symptom and they identify the code causing the issue (and recommend a fix). They’re fairly methodical, and often ask me to run diagnostic code (or do it themselves).
RA_Fisher··on Show HN: I made a spreadsheet where formulas also update backwards
Thanks for sharing. It’s an alluring idea. I think you’d find the concept of statistical identifiability interesting.
RA_Fisher··on Cloudflare outage on December 5, 2025
As a reliability statistician (and web user!), I'd love to see Cloudflare investing in reliability statistics. :)
RA_Fisher··on Transparent leadership beats servant leadership
Excellent analysis. It’s like: servant of whom (themselves)?
RA_Fisher··on Credit report shows Meta keeping $27B off its books through advanced geometry
They can get a better interest rate by using a specialty data center lender.
RA_Fisher··on AGI is not possible even in 10 years
Smart is being able to produce knowledge quickly. I’m not sure how it could be denied that AIs are capable of producing knowledge quickly (obviously extremely quickly).
RA_Fisher··on AGI is not possible even in 10 years
Yeah, I think a lot of people are very insecure. I’m genuinely sorry for them. I think the best thing to do is to derive utility from AI (to mitigate the costs).
RA_Fisher··on AGI is not possible even in 10 years
They’re urgently being given agency by companies and people.
RA_Fisher··on AGI is not possible even in 10 years
When models become sufficiently sophisticated, they practically become the phenomenon they’re modeling (by limit).
RA_Fisher··on AGI is not possible even in 10 years
Sorry about the frustration. I agree they’re far from perfect (like us). They have habits, bc they model us.
RA_Fisher··on AGI is not possible even in 10 years
> super specialist in a specific science

I’m a super specialist in statistics and GPT5 and Gemini know much more than me about the topic.

RA_Fisher··on AGI is not possible even in 10 years
> If this is true then almost all of the jobs in the world would already be replaced by AI.

We have to account for human inertia. With us people, very little completely changes over night.

RA_Fisher··on AGI is not possible even in 10 years
The paid models are already smarter than the vast majority of people.
RA_Fisher··on Python is not a great language for data science
R’s the best, bc it’s been a statistical analysis language from the beginning in 1974 (and was built and developed for the purpose of analysis / modeling). Also, the tidyverse is marvelous. It provides major productivity in organizing and augmenting the data. Then there’s ggplot, the undisputed best graphical visualization system + built-ins like barplot(), or plot().

But ultimately data analysis is going beyond Python and R into the realm of Stan and PyMC3, probabilistic programming languages. It’s because we want to do nested integrals and those software ecosystems provide the best way to do it (among other probabilistic programming languages). They allow us to understand complex situations and make good / valuable decisions.

RA_Fisher··on Pozsar's Bretton Woods III: The Framework
Thank you. If it's for strategic autonomy, it's a bad trade. China needs domestic consumption like you mention, to absorb their supply / keep people working. Weakening USD would make Chinese exports more expensive, esp. if they convert it to Yuan. Meanwhile there's a property crisis and they can stimulate demand, including for strategic autonomy. One devious possibility is just giving it to Russia. Repatriating it as Yuan to stimulate doesn't make sense to me.
RA_Fisher··on Students fight back over course taught by AI
Will make it a lot cheaper and more accessible.
RA_Fisher··on Pozsar's Bretton Woods III: The Framework
Good points, I was wrong. Thank you for your thoughtful reply. Given that, I’d expect USDYUAN to rise and further reduction in China’s GDP growth.
RA_Fisher··on Measuring political bias in Claude
Exactly, I’d expect reducing its judgement has spillover effects, bc in a sense everything is political. ie- the idea of making it wise and unwise at the same time is incoherent. Bias comes at the expense of information.
RA_Fisher··on Pozsar's Bretton Woods III: The Framework
China has no serious alternative to Treasuries (bc the closest competitors have lower yields). Also, other countries don’t place themselves in the same group as Russia, bc they violated international norms in an extreme manner.
RA_Fisher··on Pozsar's Bretton Woods III: The Framework
Pozsar assigns far too much weight to Russia. It’s an economically small country. He should be focused a lot more on large American companies that carry actual heft.
RA_Fisher··on Caffeinated coffee consumption or abstinence to reduce atrial fibrillation
There’s an average amount in the average cup of coffee.
RA_Fisher··on Stochastic computing
Crickets, crickets, crickets ... :)
RA_Fisher··on Stochastic computing
How is this not rediscovering statistics in unprincipled ways?
RA_Fisher··on These Men dove to the Edmund Fitzgerald shipwreck decades ago. Their stories
Thank you. :) I try to be careful with my claims, but when I find a big one I don’t shy away from it.

Yeah, reducing patriarchal arrogance is a major opportunity for improvement in our societies.

RA_Fisher··on These Men dove to the Edmund Fitzgerald shipwreck decades ago. Their stories
> On the fateful evening of November 10, 1975, McSorley reported he had never seen bigger seas in his life.[70] Paquette, captain of Wilfred Sykes, out in the same storm, said, "I'll tell anyone that it was a monster sea washing solid water over the deck of every vessel out there."[172] The USCG did not broadcast that all ships should seek safe anchorage until after 3:35 p.m. on November 10, many hours after the weather was upgraded from a gale to a storm.[55] McSorley was known as a "heavy weather captain"[173] who, according to George Burgner, "'beat hell' out of the Fitzgerald and 'very seldom ever hauled up for weather'".[140] Paquette held the opinion that negligence caused Edmund Fitzgerald to founder. He said, "in my opinion, all the subsequent events arose because (McSorley) kept pushing that ship and didn't have enough training in weather forecasting to use common sense and pick a route out of the worst of the wind and seas."

https://en.wikipedia.org/wiki/SS_Edmund_Fitzgerald

← PreviousPage 3 of 19Next →