Emily Hunt
@emily.space
Postdoc @univie.ac.at
Researches the Milky Way & star clusters with machine learning
Founded the Astronomy feeds (@astronomy.blue)
π³οΈβπ π³οΈββ§οΈ (she/her), β
Website: https://emily.space
GitHub: https://github.com/emilyhunt
Researches the Milky Way & star clusters with machine learning
Founded the Astronomy feeds (@astronomy.blue)
π³οΈβπ π³οΈββ§οΈ (she/her), β
Website: https://emily.space
GitHub: https://github.com/emilyhunt
Pinned
Emily Hunt
@emily.space
Β· Oct 16
If you like astronomy posts, then you'll probably really like the astronomy feeds!
If you didn't know, Bluesky gives you direct control over your social algorithm. Instead of doomscrolling, look at space instead:
If you didn't know, Bluesky gives you direct control over your social algorithm. Instead of doomscrolling, look at space instead:
Reposted by Emily Hunt
Gender-affirming care saves lives
1. A landmark study was just published in The Journal of Pediatrics.
It found a 68% reduction in suicidality for trans youth getting HRT.
It also found only 7 of more than 400 stopped taking HRT... and of those that did, 4 still identified as gender-diverse.
Transgender care saves lives.
It found a 68% reduction in suicidality for trans youth getting HRT.
It also found only 7 of more than 400 stopped taking HRT... and of those that did, 4 still identified as gender-diverse.
Transgender care saves lives.
Study In The Journal Of Pediatrics Finds Trans Youth Care Lowers Suicidality, Few Detransition
The groundbreaking study found that suicidality dropped for transgender youth receiving hormone therapy by nearly 70%, with only 7 patients of 432 discontinuing treatment.
www.erininthemorning.com
November 25, 2025 at 11:49 AM
Gender-affirming care saves lives
SAP software is a scourge on humanity
November 25, 2025 at 9:58 AM
SAP software is a scourge on humanity
Reposted by Emily Hunt
π Comet Lemmon and the Milky Way
Image Credit & Copyright: Lin Zixuan linzx23 at mails.tsinghua.edu dot cn) (Tsinghua U.)
apod.nasa.gov/apod/ap25112...
Image Credit & Copyright: Lin Zixuan linzx23 at mails.tsinghua.edu dot cn) (Tsinghua U.)
apod.nasa.gov/apod/ap25112...
November 25, 2025 at 8:00 AM
π Comet Lemmon and the Milky Way
Image Credit & Copyright: Lin Zixuan linzx23 at mails.tsinghua.edu dot cn) (Tsinghua U.)
apod.nasa.gov/apod/ap25112...
Image Credit & Copyright: Lin Zixuan linzx23 at mails.tsinghua.edu dot cn) (Tsinghua U.)
apod.nasa.gov/apod/ap25112...
It's especially dumb because a lot of those unis make a net loss on their domestic students (and prop up their finances with international fees.) Not to mention that having a big international student sector ought to be a huge soft power tool for the UK. But alas
November 24, 2025 at 8:59 AM
It's especially dumb because a lot of those unis make a net loss on their domestic students (and prop up their finances with international fees.) Not to mention that having a big international student sector ought to be a huge soft power tool for the UK. But alas
Sorry for my cynicism π
thank you for the work you do and it's good that where you work seems to do it better! I agree in principle, science communication should be a job for science communicators
November 23, 2025 at 11:56 AM
Sorry for my cynicism π
thank you for the work you do and it's good that where you work seems to do it better! I agree in principle, science communication should be a job for science communicators
Trouble is, most of us don't have one, or if they do they ask us to write it ourselves anyway... and: their priorities are warped (winning = get as much attention for the uni as possible) so I don't think comms teams would even necessarily be better π«€
November 23, 2025 at 11:12 AM
Trouble is, most of us don't have one, or if they do they ask us to write it ourselves anyway... and: their priorities are warped (winning = get as much attention for the uni as possible) so I don't think comms teams would even necessarily be better π«€
It's awfully convenient to suddenly start calling everything AI the second it becomes a buzzword, though.
And as science communicators, the minutiae of which field something is in is completely irrelevant. Intentionally using a buzzword in a way the public will misinterpret isn't helpful.
And as science communicators, the minutiae of which field something is in is completely irrelevant. Intentionally using a buzzword in a way the public will misinterpret isn't helpful.
November 23, 2025 at 11:10 AM
It's awfully convenient to suddenly start calling everything AI the second it becomes a buzzword, though.
And as science communicators, the minutiae of which field something is in is completely irrelevant. Intentionally using a buzzword in a way the public will misinterpret isn't helpful.
And as science communicators, the minutiae of which field something is in is completely irrelevant. Intentionally using a buzzword in a way the public will misinterpret isn't helpful.
Call it AI if you somehow make an LLM discover actual new physics (and please not just something it scraped off of the internet.)
This will probably not happen for a few more decades at best... we are probably far from AGI, and anyone telling you otherwise is trying to raise their share price.
This will probably not happen for a few more decades at best... we are probably far from AGI, and anyone telling you otherwise is trying to raise their share price.
November 23, 2025 at 9:24 AM
Call it AI if you somehow make an LLM discover actual new physics (and please not just something it scraped off of the internet.)
This will probably not happen for a few more decades at best... we are probably far from AGI, and anyone telling you otherwise is trying to raise their share price.
This will probably not happen for a few more decades at best... we are probably far from AGI, and anyone telling you otherwise is trying to raise their share price.
My family are useful canaries of what news the public actually sees from astro, and only yesterday did my mum ask me about a study (one article of many linked below) who said they "trained an AI" to simulate the Milky Way.
This was NOT AI as how the public sees "AI", and this is disingenuous.
This was NOT AI as how the public sees "AI", and this is disingenuous.
November 23, 2025 at 9:21 AM
My family are useful canaries of what news the public actually sees from astro, and only yesterday did my mum ask me about a study (one article of many linked below) who said they "trained an AI" to simulate the Milky Way.
This was NOT AI as how the public sees "AI", and this is disingenuous.
This was NOT AI as how the public sees "AI", and this is disingenuous.
Reposted by Emily Hunt
Machine learning (umbrella term, I know) is a useful, sometimes transformative tool in the hands of trained researchers who understand how to deploy it and critically assess the results.
A chatbot is not useful in the same ways (though underlying technologies may be, in other contexts).
A chatbot is not useful in the same ways (though underlying technologies may be, in other contexts).
November 22, 2025 at 5:01 PM
Machine learning (umbrella term, I know) is a useful, sometimes transformative tool in the hands of trained researchers who understand how to deploy it and critically assess the results.
A chatbot is not useful in the same ways (though underlying technologies may be, in other contexts).
A chatbot is not useful in the same ways (though underlying technologies may be, in other contexts).
Reposted by Emily Hunt
Lumping them together as βAIβ gives readers the impression that a single class of tool is discerning novel protein structures, teasing subtle patterns out of mountains of LHC data, writing a studentβs History 101 paper, and arguing a ketamine-addled billionaire could post up Shaq in his prime. No.
November 22, 2025 at 5:10 PM
Lumping them together as βAIβ gives readers the impression that a single class of tool is discerning novel protein structures, teasing subtle patterns out of mountains of LHC data, writing a studentβs History 101 paper, and arguing a ketamine-addled billionaire could post up Shaq in his prime. No.
Reposted by Emily Hunt
Anyway, do not help the Sam Altmans of the world steal Machine Learning valor to prop up a bunch of gross plagiarism machines that have all our worst biases and failings baked in.
November 22, 2025 at 5:14 PM
Anyway, do not help the Sam Altmans of the world steal Machine Learning valor to prop up a bunch of gross plagiarism machines that have all our worst biases and failings baked in.
PLEASE this.
And also to scientists writing press releases, too - calling something "AI" when it was actually your student spending 12 months fitting and validating a model is disingenuous
And also to scientists writing press releases, too - calling something "AI" when it was actually your student spending 12 months fitting and validating a model is disingenuous
Journalist challenge: Use βMachine Learningβ when you mean machine learning and βLLMβ when you mean LLM. Ditch βAIβ as a catch-all term, itβs not useful for readers and it helps companies trying to confuse the public by obscuring the roles played by different technologies. π§ͺ
November 23, 2025 at 9:12 AM
PLEASE this.
And also to scientists writing press releases, too - calling something "AI" when it was actually your student spending 12 months fitting and validating a model is disingenuous
And also to scientists writing press releases, too - calling something "AI" when it was actually your student spending 12 months fitting and validating a model is disingenuous
Seems to be this simple, yeah. Ours is correct, which implies it's a Bluesky issue (unless Python somehow isn't getting the right system time.)
November 21, 2025 at 1:44 PM
Seems to be this simple, yeah. Ours is correct, which implies it's a Bluesky issue (unless Python somehow isn't getting the right system time.)
For more info: this seems to effect a small minority of users (~1%) but quite consistently. Quite a few people couldn't see the feed at all, at all times, due to this issue.
November 21, 2025 at 10:19 AM
For more info: this seems to effect a small minority of users (~1%) but quite consistently. Quite a few people couldn't see the feed at all, at all times, due to this issue.
Custom feed maintainers who have authenticated feeds: has anyone else started having issues recently with immature json web tokens (JWTs) affecting certain users?
At least ~2 feeds using the Python SDK have encountered it, but we aren't sure if it's exclusive to that or a wider issue
At least ~2 feeds using the Python SDK have encountered it, but we aren't sure if it's exclusive to that or a wider issue
Occasional "TokenImmatureSignatureError" on validating JWTs with authenticated feed Β· Issue #38 Β· MarshalX/bluesky-feed-generator
Hi, for at least a week we've been repeatedly seeing this error on the Astronomy feeds Flask server backend (which closely follows the implementation in this library), where some users' DID cannot ...
github.com
November 21, 2025 at 10:19 AM
Custom feed maintainers who have authenticated feeds: has anyone else started having issues recently with immature json web tokens (JWTs) affecting certain users?
At least ~2 feeds using the Python SDK have encountered it, but we aren't sure if it's exclusive to that or a wider issue
At least ~2 feeds using the Python SDK have encountered it, but we aren't sure if it's exclusive to that or a wider issue
Has been fun following #EuroskyLive updates today! Sad to not be there in person (promised myself I'd not travel too much more after moving country this summer π
), hope to get a chance to meet other atproto devs in person someday soon
November 19, 2025 at 6:32 PM
Has been fun following #EuroskyLive updates today! Sad to not be there in person (promised myself I'd not travel too much more after moving country this summer π
), hope to get a chance to meet other atproto devs in person someday soon
It's an absolutely gorgeous and historical building that's annoyingly difficult to photograph as it sits in a lovely bit of woodland
This was it just after construction (with fewer trees)
This was it just after construction (with fewer trees)
November 19, 2025 at 10:48 AM
It's an absolutely gorgeous and historical building that's annoyingly difficult to photograph as it sits in a lovely bit of woodland
This was it just after construction (with fewer trees)
This was it just after construction (with fewer trees)
golden hour at the @univie.ac.at observatory
November 19, 2025 at 10:46 AM
golden hour at the @univie.ac.at observatory
Final question - do you know roughly when it stopped working, and if you could never see the feed after that point or if it was intermittent?
The server is still having errors with some users/requests, just it's catching them now so you can at least see the feeds still.
The server is still having errors with some users/requests, just it's catching them now so you can at least see the feeds still.
November 19, 2025 at 8:15 AM
Final question - do you know roughly when it stopped working, and if you could never see the feed after that point or if it was intermittent?
The server is still having errors with some users/requests, just it's catching them now so you can at least see the feeds still.
The server is still having errors with some users/requests, just it's catching them now so you can at least see the feeds still.
These quotes from this article live rent-free in my head now: Meta saying their security is "industry-leading" when they're *eight years* behind on implementing a fix for a known issue π«
November 19, 2025 at 8:13 AM
These quotes from this article live rent-free in my head now: Meta saying their security is "industry-leading" when they're *eight years* behind on implementing a fix for a known issue π«
Reposted by Emily Hunt
A new 3D map of interstellar dust with Gaia! π
A new study, led by Marie Barbillon, exploited the spectroscopic parametriser from Gaia DR3 to build a high-resolution extinction map! It extends up to 4 kpc from the Sun, with a finer version focused on the Local Bubble area.
#astro #galactic
A new study, led by Marie Barbillon, exploited the spectroscopic parametriser from Gaia DR3 to build a high-resolution extinction map! It extends up to 4 kpc from the Sun, with a finer version focused on the Local Bubble area.
#astro #galactic
November 18, 2025 at 9:35 AM