Zach Hensel
@zachhensel.bsky.social
single-molecule microbiology lab @ ITQB NOVA in Portugal
https://zach-hensel.github.io/
https://zach-hensel.github.io/
Does an arms race make intuitive sense here? The timescales seem very different.
November 11, 2025 at 11:23 AM
Does an arms race make intuitive sense here? The timescales seem very different.
But they’re also increasingly good at assisting with literature search… so it should’ve been trivial to find a precedent if there was one to find.
November 8, 2025 at 7:43 PM
But they’re also increasingly good at assisting with literature search… so it should’ve been trivial to find a precedent if there was one to find.
Yeah my response was that it’s very, very difficult not to encode the answer you want in your question. And LLMs are trained to tell you what you want to hear.
November 8, 2025 at 7:42 PM
Yeah my response was that it’s very, very difficult not to encode the answer you want in your question. And LLMs are trained to tell you what you want to hear.
Someone made an argument the other day, and I asked for any precedent in the literature, which one would expect to find if it was a good argument.
The eventual response was that Gemini, Chat-GPT, and Grok all agreed it was a plausible argument 🤷
The eventual response was that Gemini, Chat-GPT, and Grok all agreed it was a plausible argument 🤷
November 8, 2025 at 6:45 PM
Someone made an argument the other day, and I asked for any precedent in the literature, which one would expect to find if it was a good argument.
The eventual response was that Gemini, Chat-GPT, and Grok all agreed it was a plausible argument 🤷
The eventual response was that Gemini, Chat-GPT, and Grok all agreed it was a plausible argument 🤷
this update on the gisaid site is so weird gisaid.org/resources/st...
November 6, 2025 at 10:36 PM
this update on the gisaid site is so weird gisaid.org/resources/st...
Colonialists taking all of the harvest away, apparently.
Also given what we learned a couple decades ago, palm-civet-based agriculture might be something to avoid.
Also given what we learned a couple decades ago, palm-civet-based agriculture might be something to avoid.
November 6, 2025 at 9:39 AM
Colonialists taking all of the harvest away, apparently.
Also given what we learned a couple decades ago, palm-civet-based agriculture might be something to avoid.
Also given what we learned a couple decades ago, palm-civet-based agriculture might be something to avoid.
Thanks! It's odd because the main analysis in the paper is just a way worse version of this -- github.com/alexcritschr... -- it's rhetorically useful that all 11 sites are so obviously natural, but probably the odds were against it with my simplified method since it's noisier.
November 5, 2025 at 7:20 PM
Thanks! It's odd because the main analysis in the paper is just a way worse version of this -- github.com/alexcritschr... -- it's rhetorically useful that all 11 sites are so obviously natural, but probably the odds were against it with my simplified method since it's noisier.
This was obviously false when published... I'm most interested in learning how it was influential in spite of this and if it's possible to prevent a repeat in the future.
November 5, 2025 at 7:02 PM
This was obviously false when published... I'm most interested in learning how it was influential in spite of this and if it's possible to prevent a repeat in the future.
It wouldn't be so tough to produce these figures for each election and update quickly, but you'd have to be ready each election to handle whatever secretaries of state decide to change and you'd be lagging the AP data everyone pays for.
November 5, 2025 at 3:42 PM
It wouldn't be so tough to produce these figures for each election and update quickly, but you'd have to be ready each election to handle whatever secretaries of state decide to change and you'd be lagging the AP data everyone pays for.
I forgot how well McMullin did in Utah
November 5, 2025 at 3:35 PM
I forgot how well McMullin did in Utah
Quick search and can't find any free source for results data in general e.g., scraping the Secretary of State pages.
November 5, 2025 at 3:26 PM
Quick search and can't find any free source for results data in general e.g., scraping the Secretary of State pages.
The other way to do it that's less intuitively confusing for people is to center circles on each jurisdiction scaled by population size and expand them until they can't be expanded anymore... but I think we have worse intuition for relative areas of circles than relative areas of random weird shapes
November 5, 2025 at 3:18 PM
The other way to do it that's less intuitively confusing for people is to center circles on each jurisdiction scaled by population size and expand them until they can't be expanded anymore... but I think we have worse intuition for relative areas of circles than relative areas of random weird shapes
In that picture I think there's a white spot where Bedford City is since it is no longer independent... matching jurisdiction names was the only tricky part and I'm sure there's a good data source out there to avoid that.
November 5, 2025 at 3:16 PM
In that picture I think there's a white spot where Bedford City is since it is no longer independent... matching jurisdiction names was the only tricky part and I'm sure there's a good data source out there to avoid that.
One of those "I know this is possible and not so difficult but I don't know how to do it and do not care to learn and remember" things where LLMs are the way to go.
November 5, 2025 at 3:16 PM
One of those "I know this is possible and not so difficult but I don't know how to do it and do not care to learn and remember" things where LLMs are the way to go.
I scraped the vote totals from the VA site and got a VA county shapefile from whatever popped up in a Google search -- so not easily reproducible but here's how it got put together: colab.research.google.com/drive/14ZF38...
November 5, 2025 at 3:16 PM
I scraped the vote totals from the VA site and got a VA county shapefile from whatever popped up in a Google search -- so not easily reproducible but here's how it got put together: colab.research.google.com/drive/14ZF38...
The HKU5 is a good example about how you can't plan out ahead of time exactly how your restriction-ligation plan will go. There are other options without the same sequence constraints.
November 4, 2025 at 9:42 AM
The HKU5 is a good example about how you can't plan out ahead of time exactly how your restriction-ligation plan will go. There are other options without the same sequence constraints.
Quay's BsmBI/BsaI theory is that sites were removed; inconvenient sites not appearing in RecCA shows that this was wrong, too.
November 4, 2025 at 9:40 AM
Quay's BsmBI/BsaI theory is that sites were removed; inconvenient sites not appearing in RecCA shows that this was wrong, too.
@mbweissman.bsky.social btw, the first of these is accurately described as "seven contiguous cDNA pieces linked by unique restriction endonuclease sites that do not disturb the coding sequence" in that the cDNA pieces have synonymous mutations.
November 4, 2025 at 9:40 AM
@mbweissman.bsky.social btw, the first of these is accurately described as "seven contiguous cDNA pieces linked by unique restriction endonuclease sites that do not disturb the coding sequence" in that the cDNA pieces have synonymous mutations.
And none of that is from Pekar 2022, but whatever.
November 3, 2025 at 11:36 PM
And none of that is from Pekar 2022, but whatever.
Feel free to repeat the analysis yourself... or just look at the data and think about it and you can approximate the probabilities well enough.
For sure, C29095T happened pretty early. Not impossible that it's ancestral, but I'd bet on another low likelihood scenario over that one.
For sure, C29095T happened pretty early. Not impossible that it's ancestral, but I'd bet on another low likelihood scenario over that one.
November 3, 2025 at 11:35 PM
Feel free to repeat the analysis yourself... or just look at the data and think about it and you can approximate the probabilities well enough.
For sure, C29095T happened pretty early. Not impossible that it's ancestral, but I'd bet on another low likelihood scenario over that one.
For sure, C29095T happened pretty early. Not impossible that it's ancestral, but I'd bet on another low likelihood scenario over that one.
For sure, if this is what you read, I can understand thinking that our manuscript can be accurately summarized from picking one sentence out of the fourth paragraph of our discussion and misrepresenting what we wrote, though!
November 3, 2025 at 11:24 PM
For sure, if this is what you read, I can understand thinking that our manuscript can be accurately summarized from picking one sentence out of the fourth paragraph of our discussion and misrepresenting what we wrote, though!
That's not a claim in this manuscript; it's summarizing the result of analysis from another manuscript; results from supplemental tables from www.cell.com/cell/fulltex... and academic.oup.com/ve/article/1...
Not sure what Bloom is rebutting given the first sentence of our paragraph here.
Not sure what Bloom is rebutting given the first sentence of our paragraph here.
November 3, 2025 at 11:19 PM
That's not a claim in this manuscript; it's summarizing the result of analysis from another manuscript; results from supplemental tables from www.cell.com/cell/fulltex... and academic.oup.com/ve/article/1...
Not sure what Bloom is rebutting given the first sentence of our paragraph here.
Not sure what Bloom is rebutting given the first sentence of our paragraph here.