This is a really interesting, well-written story about some recent work on elliptic curves. https://www.quantamagazine.org/elliptic-curve-murmurations-found-with-ai-take-flight-20240305/

The problem with this claim is that it’s misleading to ordinary people. The sniff test is easy: if *my* data doesn’t matter for constructing the model, why do you need it?

And the answer is that your data is essential to building the model.

What’s actually happening here is that the model relies on collecting statistical aggregate data from collections of people. It may not know that you’re a person who likes eg specific types of private media. But if there are others like you, it learns to recognize you as a group.

https://www.cbc.ca/news/canada/british-columbia/air-canada-chatbot-lawsuit-1.7116416

ht @kennwhite

Chatbot gave wrong information, Air Canada had to pay $812 in court.

My take: Some executive is getting a fat bonus for switching Air Canada support to a system with no employees that only costs them $812 in lawsuits

@futurebird @hrefna yeah, I guess I just hate the permissive types in python

@futurebird @hrefna this feels like a perfect opportunity to teach an object oriented approach. Resistance object with a "series or parallel" attribute, and an internal list of resistance objects. One method of "compute resistance" which calls "compute resistance" on elements of its list.

Though, parsing lists of lists into this object is basically the original problem

Two things happen once the 10 women are added to the team. One of the things that happens is very obvious, and it has a small effect on the score. The other thing that happens is much less obvious, but has a much larger effect on the score.

Small thing:
The 10 women do have experience with microaggressions that affect women.

Huge thing:
All 100 men on the engineering team observe how the 10 women are treated, especially the top 2 women performers.

@tiffcassidy https://mastodon.social/@RebelGeo/111756182301184464

Wikipedia has a funny definition of "sophisticated" here:

@atoponce like a cursed python fever dream where integer literals are mutable global variables

@futurebird maybe more direct to your question, I think yes, getting a beneficial mutation yourself is very unlikely, but having a beneficial mutation somewhere in the population is much more likely. So, eg, climate changes and you need to flower later, youre not going to just happen to mutate that way (nor your descendants), but maybe there were plants in your species already flowering kind of late.

@futurebird https://www.science.org/doi/10.1126/science.1194513 had an interesting study of nightshades. It absolutely benefits the individual to self fertilize, so a mutation that allows this will spread rapidly. But then the species is less able to adapt to environmental change and thus goes extinct more frequently. Nightshades speciate fast enough to maintain a big enough collection of non-self-fertilizing species.

@futurebird One way to look at it: label each term in the nth derivative by (i,j) such that the term is the ith derivative of f and the jth derivative of g (and i+j=n). Then the coefficients of these terms will form Pascals triangle, with the nth level of the triangle representing all terms in the nth derivative. Not too hard to prove with the product rule that the coefficients of (i,j) in the nth derivstive is the sum of coefficients of (i-1,j) and (i,j-1) in the n-1 derivative

Policy idea: if a company drops support for a software they develop, in 5 years they must open-source the code

@jeffcliff Who knows what will happen in the future? Microsoft's taking a high-risk, high-reward strategy with these, but even if it works they will need further error correction strategies, like everyone else.

@luca_defeo just establishing prior art so the CCS paper has to cite this toot :)

The year is 2033. A paper is just accepted to CCS which recovers thousands of Kyber secret keys in the wild, because they removed the re-encryption step during decapsulation for performance reasons

@bascule I don't see a huge crash in the data ? In any case overall energy demand seems to be climbing steadily

@bascule @sophieschmieg IEA guy once told my partner: fridges got more efficient over the last 20 years but people still use the same amount of energy for refrigeration, because they bought bigger fridges. Renewable energy doesn't help us if we can't manage renewable consumption

@dotcolm they did it without reflections! https://mathstodon.xyz/@csk/110454664118770148