@lana@mstdn.science avatar

lana

@lana@mstdn.science

Computer Science, among other things. Artificial Life, AI, Astrobiology. views my own. Research @ Sony CSL; Associate Professor @ NIBB
member of ALife board & ALifeNewsletter. talking dog insta @iikocookie

This profile is from a federated server and may be incomplete. Browse more on the original instance.

lana, to random
@lana@mstdn.science avatar

Overleaf, primarily used to write scientific papers, encouraging its users to use AI text. What could go wrong! I am even less open to reviewing papers now. Waiting for Evilsevier to release its "AI for peer reviewers" tool so that we can live in the most boring world ever for a few years. Just until the next generation, with better BS detectors than us, start doing real science again

lana, to random
@lana@mstdn.science avatar

Among all the things LLMs taught me, one is that a sizeable portion of humanity only has the vaguest grasp of their own language. Inside their heads all might be very clear and sensible, but the means by which that gets transferred between them and other people is a big old cloud of vibes, carried by words and sentences that only need to look like they make sense. That's enough to communicate with people they see everyday so they just don't notice it breaks down outside the inner circle.

lana, to random
@lana@mstdn.science avatar

Our International School in Kyoto has nice books in its library
cc @ZachWeinersmith

lana, to random
@lana@mstdn.science avatar

Oh dang! I hate writing citations and I've always used Scholar to write them for me (even for citing my own papers, and it never got that wrong). Nothing to do with wether I've read the papers...
Also despite many attempts over the years I've never managed to make zotero generate latex citations....
https://dair-community.social/@alex/112497411007577712

lana, to random
@lana@mstdn.science avatar

Since you're not supposed to eat grapefruit while taking meds as it enhances the effects of some meds to dangerous levels, why aren't those meds sold in smaller amounts premixed with grapefruit?

lana, to random
@lana@mstdn.science avatar

I am still looking for a student (PhD or Master) in Japan who would like a side job in web development. ~8 hours per week, ~1500 to 1800 y/hour depending on your current university grade.

Contact me lana.sinapayen@gmail.com

neuralreckoning, to science
@neuralreckoning@neuromatch.social avatar

Thought about hypothesis testing as an approach to doing science. Not sure if new, would be interested if it's already been discussed. Basically, hypothesis testing is inefficient because you can only get 1 bit of information per experiment at most.

In practice, much less on average. If the hypothesis is not rejected you get close to 0 bits, and if it is rejected it's not even 1 bit because there's a chance the experiment is wrong.

One way to think about this is error signals. In machine learning we do much better if we can have a gradient than just a correct/false signal. How do you design science to maximise the information content of the error signal?

In modelling I think you can partly do that by conducting detailed parameters sweeps and model comparisons. More generally, I think you want to maximise the gain in "understanding" the model behaviour, in some sense.

This is very different to using a model to fit existing data (0 bits per study) or make a prediction (at most 1 bit per model+experiment). I think it might be more compatible with thinking of modelling as conceptual play.

I feel like both experimentalists and modellers do this when given the freedom to do so, but when they impose a particular philosophy of hypothesis testing on each other (grant and publication review), this gets lost.

Incidentally this is also exactly the problem with our traditional publication system that only gives you 1 bit of information about a paper (that it was accepted), rather than giving a richer, open system of peer feedback.

lana,
@lana@mstdn.science avatar

@neuralreckoning 2 proposals:

  1. Choose hypotheses where both possibilities are exciting (i feel like a lot of null hypotheses are borderline stupid, and that makes them bad null hypotheses. If you look at things like engineering in space, when something goes wrong in a spacecraft and you only have a 2 min window of communication with 2 days to get an answer... people find way to maximize the info out of their null hypotheses)
  2. Use Taguchi arrays
    https://youtu.be/5oULEuOoRd0?si=CZnpXkgwpuPphN5Y
lana, to random
@lana@mstdn.science avatar

Dragonfly is an octocopter that will explore Titan, the ocean-world moon that has ice and liquid water and complex molecules that could be precursors of life (or even early life)

QR code to an AR version of Dragonfly

neuralreckoning, to academia
@neuralreckoning@neuromatch.social avatar

So oral exam at end of PhD. Good idea or just a tradition that doesn't make any sense any more? What are the good things about them? If we didn't do them, how else could we get those good things?

lana,
@lana@mstdn.science avatar

@neuralreckoning I've seen it be a good 'rite of passage' for the student, in the sense that it feels like a more fitting closure than just the moment you hand out the file.
You get the sense of trepidation and release, often followed by a celebratory dinner with your lab, that you don't get from doing the actual important parts (the years of research and months of writing)

lana, to random
@lana@mstdn.science avatar

https://www.cell.com/iscience/fulltext/S2589-0042(24)00836-8

I really thought this was from the Levin lab!

lana,
@lana@mstdn.science avatar

Friendlier version

https://phys.org/news/2024-04-cells-hidden-communication.html

lana, to random
@lana@mstdn.science avatar

Exceedingly cool results.
Somehow, wherever this small bat thrives, its presence also creates a niche for a bigger bat, which evolves from... the small bat. You get a string of islands with the simultaneous presence of the small bat and the big bat that re-evolves from it every time....

The phenomenon is so weird that people thought it was just 2 different species that colonized the islands in parallel. It's just one species with a recurring sub-branch.

https://phys.org/news/2024-04-species-sizes-rare-evolution-action.html

lana, to random
@lana@mstdn.science avatar

A lot of media is leaving out the main point of our paper and cheapening our results :/

The main point isn't simply "look for a bunch of planets that look the same", it's that these planets would DIFFER in a predictable, highly nonrandom way indicating that life is mutating planet by planet, and adapting measures from ecological research we can even tell you how unlikely that arrangement would be: 1 in a thousand, a million, billion?

https://www.popularmechanics.com/space/deep-space/a60505786/alien-world-similar-planet/

lana, to random
@lana@mstdn.science avatar

The whole thread is very informative but the most important thing I learned is that in this day and age even IT companies believe chatbots will understand and obey instructions if you just say please 🙄
People who should know better have gaslighted themselves into believing they have truly built Artificial Intelligence.

https://eldritch.cafe/@Lugrim/112258207338842046

lana, to random
@lana@mstdn.science avatar

We're on phys.org
https://phys.org/news/2024-03-life-difficult-planet-obvious-worlds.html

And in Spanish:
https://www.astrobitacora.com/la-vida-podria-ser-facil-de-detectar-si-se-observan-muchos-mundos/

lana, to random
@lana@mstdn.science avatar

Here it is! Our paper about a new way to detect life in the universe. I'll make a fediprint thread later!

"An Agnostic Biosignature Based on Modeling Panspermia and Terraformation"
https://arxiv.org/abs/2403.14195
html: https://arxiv.org/html/2403.14195v1

lana, to random
@lana@mstdn.science avatar

Something came in the mail the other day 👀
In pandemic conscious academia days, in 2022, I couldn't receive it in person

NicoleCRust, to random
@NicoleCRust@neuromatch.social avatar

How do you pack a 95K word (nonfiction) book into a 40 minute talk?

How many words are in 40 minutes? My estimate is 4-5K. That's ~20-fold compression. Something like half of 1 (of 10) chapters in the book.

Obviously you don't just read off the first half of the first chapter. But an outline of all of it is also super unsatisfying; it needs more depth than that. Clearly you present the central thesis and why it matters. But what to support it? This is a problem I've never encountered before. Not yet sure how to wrap my head around it.

Any advice? Any pointers to book talks you love?

lana,
@lana@mstdn.science avatar

@NicoleCRust I think I'd take just one "case study" or example that makes the point of the whole book if possible, and go narratively through that example keeping only the essential parts of the argument. If after that I noticed there is still time, I might add a second one, or more likely talk about something that is not in the book: some behind the scenes, stuff that didn't make the cut, how did I decide to write about this, what are the reactions to the book so far, any interviews

lana, to random
@lana@mstdn.science avatar

Winner of Dance Your PhD 2024

And also coolest person on earth

https://youtu.be/RoSYO3fApEc?si=dBgE-OezNwoji4lZ

lana, to random
@lana@mstdn.science avatar

Did I just get old or are memes just not a thing anymore? No more images or story templates reused to death, now it's only viral videos and 5 reaction gifs

lana, to random
@lana@mstdn.science avatar

About the genocide in Palestine.

https://www.youtube.com/watch?v=3xottY-7m3k&ab_channel=Shaun

FlockOfCats, (edited ) to random
@FlockOfCats@famichiki.jp avatar

Remember when the Japan government sent everyone two (!) masks (Abe no mask), but it turned out many of them were defective?

They apparently still have extras laying around, and they use them at a Tokyo kid’s center as name tags. 😳

The kids’ names are stitched on and they wear the mask on their back :gosh:

lana,
@lana@mstdn.science avatar

@FlockOfCats Looks like the straps are used to keep it in place, rather than a stitch. Do you still have the link to the original source? I'm curious now

lana, to random
@lana@mstdn.science avatar

Anyone in Japan who would be eager to share advice about moving to Japan with a cat? Can I put you in contact with the person needing advice? I can give them tips about the city they're going to live in, but I don't have cat expertise. All I know is it's even harder to rent than with a dog.

lana, to random
@lana@mstdn.science avatar

Paper purporting to have fascinating results.
Paywalled. It's 2024 people! How is this still a thing! Can you downvote a website??

https://www.nature.com/articles/s41562-024-01815-w

lana,
@lana@mstdn.science avatar

Shamefully, I have one paywalled Nature paper. We all agreed in advance that we would make it open access (I'm not 1st author). Come publication day, I notice that it's not open access. 1st author checks with editor. Editor says it's "too late" to make it open access now that it's been published, even if it was published paywalled by mistake. These people still live in 1924. (we put the paper on Arxiv.)

  • All
  • Subscribed
  • Moderated
  • Favorites
  • •
  • provamag3
  • InstantRegret
  • mdbf
  • ethstaker
  • magazineikmin
  • GTA5RPClips
  • rosin
  • thenastyranch
  • Youngstown
  • osvaldo12
  • slotface
  • khanakhh
  • kavyap
  • DreamBathrooms
  • JUstTest
  • Durango
  • everett
  • cisconetworking
  • Leos
  • normalnudes
  • cubers
  • modclub
  • ngwrru68w68
  • tacticalgear
  • megavids
  • anitta
  • tester
  • lostlight
  • All magazines
    •