lindenr

I agree, if AI (or humans) have mistranslated a natural language statement to a formal statement, we should not rely on AI to correctly translate the formal statement back into natural language.

For many statements I expect it's not possible to retain the exact meaning of the formal-language sentence without the natural language becoming at least as complex, and if you don't retain meaning exactly then you're vulnerable to the kind of thing the article warns about.

Really interesting! It does need an explanation - took me a bit of clicking around to discover the rules (which I actually enjoyed because it turned out not to be too complicated! But some will give up early).

One aspect I liked about the larger puzzle sizes it that the topology becomes nontrivial: you can work out for sure that a cell has to be marked without knowing the exact path from the edge inwards, which you later narrow down.

Feature request: can interior cells be marked? Right now it's disallowed which helps rule discovery but requires the user to remember more.

Yeah, I think it's a typo and should have been "one in every 24 million babies" - this makes sense given the numbers, with 4 million born every year but only a new president every 6 years.

If I understand you right - you could visualize in 2d space: "king" at origin, X-axis is "king"-"man", Y-axis is "king"-"woman" (or gram-schmidt if you really want orthogonal).

In 3d you can go one further and have the Z-axis be "king"-"queen" (or gram-schmidt again). The orthogonalized versions have the advantage that they give a closer notion of distance to what the underlying model sees. In the 2d case you will get exact distances except that it won't show how far off "queen" you are when you compute "king"-"man"+"woman". In the 3d case it should give exact distances.

Edit to add: With the 2d version you can maybe do some more stuff. IIRC "queen" is chosen as it's the word with the closest embedding to X="king"-"man"+"woman". You can put the next few closest words on the 2d chart as well, each labeled with the orthogonal distance from the 2d plane. So then "queen" should be the word with the smallest (squared distance from X) + (squared orthogonal distance from plane), which you might be able to eyeball.

Would you be interested in working in quantitative finance in London? We're hiring (remotely for now, but you'd be expected to work in London probably later this year).

We have lots of PhDs (physics, maths, etc - not exclusively though, I only have a master's in maths) and aren't particularly looking for a finance background, so you're pretty much a classic interview candidate. (Finance knowledge or interest is obviously a plus.) Interviews would be on a combination of general maths, stats, finance, ML and programming - whatever subset you're strongest in.

To sell the role a bit - we work on algorithmic trading, but I probably can't say too much more than that. I'm really enjoying working here, the environment's a lot like hanging out at uni with my nerdy friends and the work is interesting and free of bullshit. Project turnaround time is quick.

Please let me know if you'd be interested in hearing more about this. You can reach me at [email protected], but instead of my_email it's dax.

Just spent some time playing around with that, really cool! Thank you :)

Off-topic but, was that supposed to be a haiku? I only count six syllables in the second line.

You said:

"Saying "selection pressure" implies intent, determination, a collective mission, if you will. Evolution has none of those drivers. It has no objectives. It sets no goals. And it certainly does not operate in some collective purpose driven form within a population."

danieltillett said:

"At the moment the speed of disease progression to death is so fast that it is slowing the spread of Ebola. If the Ebola took longer to kill it would infect more people."

This seems like a true statement and in line with your own opinion; I can't quite see what you are objecting to.

Ah, I see. Could you edit your post to reflect this? As it stands it is very easy to misinterpret.

I thought it meant two programs which compute the same thing but possibly using different algorithms. If they are indistinguishable then you can't infer from the improved algorithm anything you can't get from the old one.

And who told you there are "hundreds of thousands" of people who want to blow up Americans?

But actually, even (very generously) at 100,000 brand-new prisoners for the USG to deal with it would certainly cope just fine (an increase of 4% on the current prison population). And imagine all the money which would be saved from surveillance programs if all the terrorists were dealt with!

It does seem a bit more forceful than bold to me; maybe you could have a separate 'bold' font and a 'really really bold' font?