I never heard the “Bernoulli” part, new?

Yeah it was named a few years ago after a student/faculty poll https://tech.caltech.edu/2023/05/update-on-bernoulli-the-bea...

FWIW I have heard the exact opposite. The people I know love the self driving feature.

This is not about loving the feature. It is about willing to pay the current asking price for it.

Agree 100% -- wonderful history here. And with such a simple statement, hard to believe / accept the "reason" is that it works for 633 limiting cases.

As somebody noticed, we're very lucky to live in a universe where many key things are utterly simple (like basic arithemetics), or allow for acceptable simple approximations (like classical mechanics). There was a sci-fi story set in a world orbiting a binary star, where discovering the laws of celestial mechanics feels like an intractable problem, and even predicting seasons is barely possible.

The privacy concerns are real.

The need / demand for some verification system might be growing though as I’ve heard fraudulent job application (people applying for jobs using fake identities… for whatever reason) is a growing trend.

Here's the wikipedia on that one

https://en.wikipedia.org/wiki/Berkson%27s_paradox

Many times I've looked at the output of a regression model, seen this effect, and then thought my model must be very bad. But then remember the points made elsewhere in thread.

One way to visually check that the fit line has the right slope is to (1) pick some x value, and then (2) ensure that the noise on top of the fit is roughly balanced on either side. I.e., that the result does look like y = prediction(x) + epsilon, with epsilon some symmetric noise.

One other point is that if you try to simulate some data as, say

y = 1.5 * x + random noise

then do a least squares fit, you will recover the 1.5 slope, and still it may look visually off to you.

Maybe comparing plots of residuals makes it clearest.

I would certainly buy a copy were it printed!

Looks like a treasure trove of knowledge.

I just recently went to the exploratorium in SF and saw an exhibit there suggesting that the catenary made a good arch, so browsed that chapter and saw a bit of explanation here which helped. Was also interested to see that Jefferson played some part in the history here.

I agree, and would add that there are others who are decidedly "anti-car" and you could say that this is part of their identity. This particular policy may be a strictly positive (no strong opinion here), but when viewed as part of the broader disagreement it drives some of the reflexive pushback.

Are you suggesting use the clip embedding for the text as a feature to train a standard Ml model on?

I think they're suggesting doing that with BERT for text and CLIP for images. Which in my experience is indeed quite effective (and easy/fast).

There have been some developments in the image-of-text/other-than-photograph area though recently. From Meta (although they seem unsure of what exactly their AI division is called): https://arxiv.org/abs/2510.05014 and Qihoo360: https://arxiv.org/abs/2510.27350 for instance.

I think he is. I do things like that plenty.