@MetaEd How did you cheat?

So that number 4 crushed me.

Of course I was in doubt between two locations.

The year was hard for me based on clothes.

The rest were easy enough.

Word (term?) of the day: iPhone. One of those odd words where a morpheme boundary inhibits (the American version of) Canadian raising; the first syllable has the vowel in lie, not the vowel in life (for myself, and I think for all speakers for whom this distinction exists).

(Compare words like hyphen, where there's no morpheme boundary and the first vowel can get raised due to the following /f/.)

Ugly marketing word.

More quotidian examples are eyesight, flypaper, bifocals, tritone, etc.

(Well, not so much more quotidian as less artificial.)

Our genome might work like Large Language Models

> .. the genome encodes a generative model of the organism. In this scheme, by analogy with variational autoencoders, the genome does not encode either organismal form or developmental processes directly, but comprises a compressed space of “latent variables”. These latent variables are the DNA sequences that specify the biochemical properties of encoded proteins and the relative affinities between trans-acting regulatory factors and their target sequence elements.

I was lucky with the years.

Only the third one I actually managed to find the year of.

@Mitch Thank you very much.

Diacritical mark of the day: diaeresis - a diacritical mark used to indicate the separation of two distinct vowels in adjacent syllables

Boötes -- /boʊˈoʊtiːz/

@jlliagre Google?

@Vikas No Google, I stopped using it a few days ago.

Only #4 has no clear hint.

I could have done a little better on #4 though because I've already been exactly there albeit not at the same time :-)

@Cerberus #4 was a cheat. Completely unfair.

@Robusto Yeah, I mean: I expected them to come to the typical people, not ship the typical people elsewhere.

Fell the beauty of lipstick

@CowperKettle I think you're putting the cart before the horse: DNA came well before LLM.

It's not me, it's the authors of hte paper

@jlliagre The rain in France stays mainly in the plain

@CowperKettle Then it's some kind of occupational myopia.

@CowperKettle Aw, just when I was ready to buy some, you spoiled it for me.

Hmm I wonder how different those really are.

@Cerberus Probably understood everywhere in der Schweiz, but a marker for identifying region.

I mean, many of those might sound almost identical.

Just spelled differently.

> In Lonesome Dove, by Larry McMurtry, a woman kills herself after living on the prairie for most of her lifetime, leaving a suicide note with the words "Tired of the damn wind".

@CowperKettle I don't remember that.

@Cerberus Great score.

@Robusto You did well yourself.

I used Google, of course.

Ah, I probably should have as well.

It is a choice.

I was proud of sussing out the location of #5.

Hmm I don't remember, what was that one?

@Cerberus SPOILER

Ah, that one.

Yeah, I guessed it must be near there, too.

Then I went on Google Maps to find it.

@Cerberus I remembered the flag of that country from watching a certain recent bicycle race.

At first I thought it was the other country, bordering it.

Then I looked up the flag when I couldn't find the city in Street View.

@Cerberus Yes, I wondered about that too. The placement was inconclusive, so I went with the first in order.

Wise.

One can understand why many people go holidaying in that region.

Yes. That whole area reminds me of the Riviera.

@CowperKettle provocative title... As a metaphor (an LLM encodes then decodes or equivalently compresses then decompresses eg can take a paragraph, compress it into latent variables, then decompress it into a paraphrase of the original text) on one side...

And then on the other side evolution takes amany individual's lives and 'compresses it' it into a genome (the latent variables of the surviving individuals) and decompresses the genome (ontogeny = development of a single individual)...

Ie gives birth (literally) to an individual...

This is a -every- tenuous analogy.

The more I think of it, it is a broken analogy. The map is: evolution is like training a neural network (nothing special about LLMs/text) and gestation after conception is like testing or inference (calculating the function on one input) in th network.

@Mitch -every- tenuous?

I think it's not such a bad analogy: both LLMs and DNA have no idea what they're doing.

Both triliteral initialisms.

The degree to which one thing is like another thing doesn't have to be accurate to six decimal places for it to be a working analogy.

What else. They both lack the letter E.

@Robusto No, indeed.

The authors are trying to explain evolution mechanisms by analogy, and I think they mean well, but in the end the analogy (if it works at all) is very tenuous. It might be useful at a very ethereal level but I don't think it will help a biologist understand thing better ,(a biologist probably doesn't have a good concept of neural networks to get benefit from an analogy

@Robusto every -> very

@Mitch I know. I was trying to give you time to correct that.

@Cerberus that's just giving them more ammunition... DNA is obviously self-conscious because we are conscious and we are made of DNA

@Robusto typing on a phone and writing an essay is hard.

QED.

@Mitch I mean, speak for yourself.

@Robusto of course. I was giving a reasonable analysis that the analogy didn't even meet the minimal 'kind of like' criterion