Deyan Ginev
@dginev.bsky.social
Software Engineer at arXiv. LaTeXML dev.
The Web was created to share science. Let's keep it that way.
creator of http://ar5iv.org
The Web was created to share science. Let's keep it that way.
creator of http://ar5iv.org
Reposted by Deyan Ginev
Wow! Yuansi Chen resolves 1 of the 2 remaining $1000 Talagrand problems (michel.talagrand.net/prizes/prize... ):
If you take any f : {-1,+1}ⁿ → ℝ⁺ and apply the noise operator T_{.99}, the resulting function g = T_{.99} f satisfies a better-than-Markov inequality. That is, Pr[g > t E[g]] < o(1/t).
If you take any f : {-1,+1}ⁿ → ℝ⁺ and apply the noise operator T_{.99}, the resulting function g = T_{.99} f satisfies a better-than-Markov inequality. That is, Pr[g > t E[g]] < o(1/t).
November 25, 2025 at 4:16 PM
Wow! Yuansi Chen resolves 1 of the 2 remaining $1000 Talagrand problems (michel.talagrand.net/prizes/prize... ):
If you take any f : {-1,+1}ⁿ → ℝ⁺ and apply the noise operator T_{.99}, the resulting function g = T_{.99} f satisfies a better-than-Markov inequality. That is, Pr[g > t E[g]] < o(1/t).
If you take any f : {-1,+1}ⁿ → ℝ⁺ and apply the noise operator T_{.99}, the resulting function g = T_{.99} f satisfies a better-than-Markov inequality. That is, Pr[g > t E[g]] < o(1/t).
Good news: you can send me complaints and feature requests.
...if they have to do with arXiv HTML Papers 🫡
...if they have to do with arXiv HTML Papers 🫡
Many people don't know this but arXiv is maintained and operated by @cornelltech.bsky.social (that does NOT mean you can send me complaints or feature requests!)
Everyone in the community should be excited that arXiv got $7M to upgrade from Schmidt Sciences, NASA
news.cornell.edu/stories/2025...
Everyone in the community should be excited that arXiv got $7M to upgrade from Schmidt Sciences, NASA
news.cornell.edu/stories/2025...
$7M grant from NASA, Schmidt Sciences to upgrade arXiv | Cornell Chronicle
The funding will help arXiv – which is maintained and operated by Cornell Tech – finish migrating to cloud infrastructure and modernizing its code.
news.cornell.edu
November 24, 2025 at 11:03 PM
Good news: you can send me complaints and feature requests.
...if they have to do with arXiv HTML Papers 🫡
...if they have to do with arXiv HTML Papers 🫡
This kind of work is also unironically the humanities jailbreaking into STEM.
November 24, 2025 at 10:04 PM
This kind of work is also unironically the humanities jailbreaking into STEM.
Serendipity:
arXiv and alphaXiv announcing identical $7 million in funds raised recently.
Key difference: non-profit donation for operations and maintenance vs for-profit ownership stake in business growth.
blog.arxiv.org/2025/11/24/a...
www.prnewswire.com/news-release...
#arXiv
arXiv and alphaXiv announcing identical $7 million in funds raised recently.
Key difference: non-profit donation for operations and maintenance vs for-profit ownership stake in business growth.
blog.arxiv.org/2025/11/24/a...
www.prnewswire.com/news-release...
#arXiv
arXiv receives $7 million in gifts and grants from Schmidt Sciences and NASA – arXiv blog
blog.arxiv.org
November 24, 2025 at 9:48 PM
Serendipity:
arXiv and alphaXiv announcing identical $7 million in funds raised recently.
Key difference: non-profit donation for operations and maintenance vs for-profit ownership stake in business growth.
blog.arxiv.org/2025/11/24/a...
www.prnewswire.com/news-release...
#arXiv
arXiv and alphaXiv announcing identical $7 million in funds raised recently.
Key difference: non-profit donation for operations and maintenance vs for-profit ownership stake in business growth.
blog.arxiv.org/2025/11/24/a...
www.prnewswire.com/news-release...
#arXiv
Reposted by Deyan Ginev
Today when trying to figure out how to write some alt-text for a professor's image of a math equation, I found out about MathML. It is so cool! A standardized, markup language that makes math equations appear correctly (and accessibly) natively in most browsers without needing LaTeX plugins.
November 24, 2025 at 8:06 PM
Today when trying to figure out how to write some alt-text for a professor's image of a math equation, I found out about MathML. It is so cool! A standardized, markup language that makes math equations appear correctly (and accessibly) natively in most browsers without needing LaTeX plugins.
Reposted by Deyan Ginev
Image screening is going to fail. We need audit trails for data provenance.
I tried an even harder example on Gemini Pro image generation and this is quite scary/amazing. I asked for a microscopy image of around 20 HeLa cells, GFP tagged 20% nuclear, 10% membrane, +1 nuclear staining, + overlap. Image below and prompt in the following post.
November 22, 2025 at 4:21 PM
Image screening is going to fail. We need audit trails for data provenance.
Reposted by Deyan Ginev
Ruining great art with the nano banana pro command “Make this much more cheerful with as few changes as possible”
November 21, 2025 at 1:19 PM
Ruining great art with the nano banana pro command “Make this much more cheerful with as few changes as possible”
Reposted by Deyan Ginev
arXiv is looking for an Engineering Director.
Help us find out what is on top of the beanstalk 🌥️
cornell.wd1.myworkdayjobs.com/en-US/Cornel...
Help us find out what is on top of the beanstalk 🌥️
cornell.wd1.myworkdayjobs.com/en-US/Cornel...
November 16, 2025 at 2:37 AM
arXiv is looking for an Engineering Director.
Help us find out what is on top of the beanstalk 🌥️
cornell.wd1.myworkdayjobs.com/en-US/Cornel...
Help us find out what is on top of the beanstalk 🌥️
cornell.wd1.myworkdayjobs.com/en-US/Cornel...
Reposted by Deyan Ginev
Reposted by Deyan Ginev
“lazy use of unwrap blew up a process and took out the internet” bzzt. wrong. judicious use of unwrap blew up a process instead of allowing an Extremely Named CVE to happen and spraying your bank account credentials all over the public internet
People want a technical solution to what is ultimately a judgement problem.
People know that unwrap can cause a panic. That's the choice that's being made when you unwrap. Changing the name won't change that.
People know that unwrap can cause a panic. That's the choice that's being made when you unwrap. Changing the name won't change that.
November 19, 2025 at 5:10 PM
“lazy use of unwrap blew up a process and took out the internet” bzzt. wrong. judicious use of unwrap blew up a process instead of allowing an Extremely Named CVE to happen and spraying your bank account credentials all over the public internet
Reposted by Deyan Ginev
🚧 OSF experienced a temporary slowdown today due to a surge of automated bot traffic. This overwhelmed parts of the production environment that didn't yet have the right safeguards in place, causing performance issues. The underlying issue has been addressed, & further protections are now in place.
November 19, 2025 at 4:00 PM
🚧 OSF experienced a temporary slowdown today due to a surge of automated bot traffic. This overwhelmed parts of the production environment that didn't yet have the right safeguards in place, causing performance issues. The underlying issue has been addressed, & further protections are now in place.
arXiv is looking for an Engineering Director.
Help us find out what is on top of the beanstalk 🌥️
cornell.wd1.myworkdayjobs.com/en-US/Cornel...
Help us find out what is on top of the beanstalk 🌥️
cornell.wd1.myworkdayjobs.com/en-US/Cornel...
November 16, 2025 at 2:37 AM
arXiv is looking for an Engineering Director.
Help us find out what is on top of the beanstalk 🌥️
cornell.wd1.myworkdayjobs.com/en-US/Cornel...
Help us find out what is on top of the beanstalk 🌥️
cornell.wd1.myworkdayjobs.com/en-US/Cornel...
Reposted by Deyan Ginev
It is as if thousands of researchers suddenly cried out in terror and were suddenly silenced
November 11, 2025 at 9:03 PM
It is as if thousands of researchers suddenly cried out in terror and were suddenly silenced
Reposted by Deyan Ginev
The package formerly known as papercheck has changed its name to metacheck! We're checking more than just papers, with functions to assess OSF projects, github repos, and AsPredicted pre-registrations, with more being developed all the time.
scienceverse.github.io/metacheck/
scienceverse.github.io/metacheck/
Check Research Outputs for Best Practices
A modular, extendable system for automatically checking research outputs for best practices using text search, R code, and/or (optional) LLM queries.
scienceverse.github.io
November 3, 2025 at 4:20 PM
The package formerly known as papercheck has changed its name to metacheck! We're checking more than just papers, with functions to assess OSF projects, github repos, and AsPredicted pre-registrations, with more being developed all the time.
scienceverse.github.io/metacheck/
scienceverse.github.io/metacheck/
🗓️ The October 2025 arXiv articles are now in ar5iv.
🗓️ The September 2025 arXiv articles are now in ar5iv.
🗓️ The August 2025 arXiv articles are now in ar5iv.
Lots of pots are brewing in September...
Lots of pots are brewing in September...
November 6, 2025 at 4:38 AM
🗓️ The October 2025 arXiv articles are now in ar5iv.
My thoughts are with everyone who continue to be affected by the government shutdown.
And I am still blocked.
And I am still blocked.
My standup blocker today was:
"I am blocked until the US Federal government reopens."
How's your day going?
"I am blocked until the US Federal government reopens."
How's your day going?
November 3, 2025 at 8:53 PM
My thoughts are with everyone who continue to be affected by the government shutdown.
And I am still blocked.
And I am still blocked.
Reposted by Deyan Ginev
JupyterBook 2 is released! This is now built on top of the @mystmd.org engine directly.
🚀
Please reach out on our GitHub discussions if you have any problems upgrading.
🚀
Please reach out on our GitHub discussions if you have any problems upgrading.
November 3, 2025 at 2:31 PM
JupyterBook 2 is released! This is now built on top of the @mystmd.org engine directly.
🚀
Please reach out on our GitHub discussions if you have any problems upgrading.
🚀
Please reach out on our GitHub discussions if you have any problems upgrading.
Reposted by Deyan Ginev
arXiv will no longer accept review articles and position papers unless they have been accepted at a journal or a conference and complete successful peer review.
This is due to being overwhelmed by a hundreds of AI generated papers a month.
Yet another open submission process killed by LLMs.
This is due to being overwhelmed by a hundreds of AI generated papers a month.
Yet another open submission process killed by LLMs.
Attention Authors: Updated Practice for Review Articles and Position Papers in arXiv CS Category – arXiv blog
blog.arxiv.org
November 1, 2025 at 5:28 PM
arXiv will no longer accept review articles and position papers unless they have been accepted at a journal or a conference and complete successful peer review.
This is due to being overwhelmed by a hundreds of AI generated papers a month.
Yet another open submission process killed by LLMs.
This is due to being overwhelmed by a hundreds of AI generated papers a month.
Yet another open submission process killed by LLMs.
Reposted by Deyan Ginev
Another casualty of the LLM revolution.
In "arXiv’s CS category, review articles and position papers must now be accepted at a journal or a conference and complete successful peer review."
In "arXiv’s CS category, review articles and position papers must now be accepted at a journal or a conference and complete successful peer review."
Attention Authors: Updated Practice for Review Articles and Position Papers in arXiv CS Category – arXiv blog
blog.arxiv.org
November 2, 2025 at 4:58 AM
Another casualty of the LLM revolution.
In "arXiv’s CS category, review articles and position papers must now be accepted at a journal or a conference and complete successful peer review."
In "arXiv’s CS category, review articles and position papers must now be accepted at a journal or a conference and complete successful peer review."
Reposted by Deyan Ginev
AI-written surveys are definitely a problem, but it feels so backwards for arXiv PREprints to require peer review first. Not sure what the solution is.
If a position paper is accepted at a decent venue (ICML, NeurIPS, ACM Computing Surveys, etc.), we will gladly release it on arXiv.
October 31, 2025 at 8:14 PM
AI-written surveys are definitely a problem, but it feels so backwards for arXiv PREprints to require peer review first. Not sure what the solution is.
Reposted by Deyan Ginev
Really inspired by the energy at the @openrxiv.bsky.social Meeting last week — feels like a turning point for preprints and open research. 🌌
Excited to see how this community is connecting data, code, and narrative into something more open, interactive, and alive. 🚀
Excited to see how this community is connecting data, code, and narrative into something more open, interactive, and alive. 🚀
The @openrxiv.bsky.social community is charting the future of preprints — not just as papers, but as connected nodes in the open research web.
Our reflections + @row1.ca's lightning talk on building continuous, connected publishing.
curvenote.com/blog/czi-ope...
Our reflections + @row1.ca's lightning talk on building continuous, connected publishing.
curvenote.com/blog/czi-ope...
October 29, 2025 at 9:15 PM
Really inspired by the energy at the @openrxiv.bsky.social Meeting last week — feels like a turning point for preprints and open research. 🌌
Excited to see how this community is connecting data, code, and narrative into something more open, interactive, and alive. 🚀
Excited to see how this community is connecting data, code, and narrative into something more open, interactive, and alive. 🚀
Reposted by Deyan Ginev
Excited to be teaming up with @creativecommons.bsky.social to think about incentives, tooling and licensing around modular scientific content.
Please sign up to join the working group!!
continuousfoundation.org/reuse-form
Please sign up to join the working group!!
continuousfoundation.org/reuse-form
@creativecommons.bsky.social and @continuous.foundation are teaming up to make reuse the default for modular science.
Join the cohort working to shape incentives for reuse in modular science. The group will kick off in January 2026.
Apply by Sunday, Nov 30
continuousfoundation.org/reuse-form
Join the cohort working to shape incentives for reuse in modular science. The group will kick off in January 2026.
Apply by Sunday, Nov 30
continuousfoundation.org/reuse-form
October 28, 2025 at 8:07 PM
Excited to be teaming up with @creativecommons.bsky.social to think about incentives, tooling and licensing around modular scientific content.
Please sign up to join the working group!!
continuousfoundation.org/reuse-form
Please sign up to join the working group!!
continuousfoundation.org/reuse-form
Reposted by Deyan Ginev
in my post-academic era i have become extremely scathing of the way academia considers papers to the primary outcome of intellectual work, and refuses to give credit to academics who write software or build other tooling. in industry we very rarely read your papers; but we always use your software
October 28, 2025 at 2:15 AM
in my post-academic era i have become extremely scathing of the way academia considers papers to the primary outcome of intellectual work, and refuses to give credit to academics who write software or build other tooling. in industry we very rarely read your papers; but we always use your software
Reposted by Deyan Ginev
