The nerdiest, and thus best, way of getting a bibitex entry from a DOI is by querying it directly via the command
curl -LH "Accept: text/bibliography; style=bibtex" doi.org/10.1016/j.ch...
The second best is just to use @zotero.org for everything.
HOO HOO HOO HOOSIERS!
Trying to watch the IU game and Sling is down. Boooooo.
Co-authors decided to pull our submission from @aclmeeting.bsky.social. None of the six of us were "qualified" reviewers, but required to register and commit to reviewing if assigned. @aclrollingreview.bsky.social were unyielding in the registration requirement.
Meh. I might pull my submission from ACL due to @aclrollingreview.bsky.social requirements. Something I feel empowered to say by anonymous submission. Will flesh out the actual commentary when I’m not on vacation.
Suggestions on alt venues for info extraction in 2026? Arxiv and call it a day?
I wrote a short essay on what I'm calling the AI Attribution Error. doi.org/10.59350/c3g...
Kind of the main point in Technofeudalism. (Highly recommended.)
In addition, the review "requirements" are growing burdensome. On a 6-author paper with 3 grad interns, 2 subject area experts, and myself, I'm the only eligible reviewer.
Will we be desk rejected when ineligible reviewers don't register?
The most frustrating submission experience I've ever had.
Anywhere on Earth, the first week of January was a terrible decision, and I can't imagine how many grad students/interns are going to spend their "holidays" trying to polish their work. Honestly, I'm only constrained by institutional review.
The @aclmeeting.bsky.social can do better.
"[T]he birth of a New Year is of an interest too wide to be pretermitted by king or cobbler. No one ever regarded the First of January with indifference. It is that from which all date their time, and count upon what is left. It is the nativity of our common Adam." - Charles Lamb
Breaking several months of Bluesky silence to tell @aclmeeting.bsky.social @aclrollingreview.bsky.social that a January 5th deadline was insanity (i.e., the first Monday of the new year).
Submitted today, December 23rd, last working day at the Labs and already a week into my break at UNM.
The Echoes in AI paper showed quite the opposite with also a story continuation setup.
Additionally, we present evidence that both *syntactic* and *discourse* diversity measures show strong homogenization that lexical and cosine used in this paper do not capture.
This is a fun genera and it is worthwhile testing to see what LLMs can’t do well. But I think Bluesky could use more posts on what the LLMs are suddenly able to do really well. For history , they can transcribe handwriting and produce historiographical overviews better than most research assistants.
My dad explaining that his grandparents in Kentucky used a washboard into the 80s...
Discussions this weekend made it clear that folks don't necessarily realize that our existing "scientific pipeline" of federal funding for basic research didn't just happen. The entire approach was deliberately designed to work this way, on purpose
Meet Vannevar Bush
some more congestion pricing Ws:
Broadway attendance: up 21%
Restaurant reservations: up 7%
Pedestrian traffic: up 4%
Retail sales: up $900M
Commercial leasing: up 61%
Subway crime: down 37%
Car crashes: down 50%
Honking complaints: down 69%
For now, I guess I boot Windows for games. Meh.
Today, I was reminded of all the joys of maintaining a Linux box: graphics card has horrid performance, possibly fixed by either upgrading the kernel (introducing a WiFi regression) or by manually updating the graphics driver, as the newest version hasn't made it to repos yet.
Things considered harmful that should not be attempted:
- Quote Posting
- Reposting
- Reply Posting
- Posting
At Perseus, the vast majority of hits have always been from non .edu addresses. NEH means the Greek and Roman world is available to everyone, not just academics.
After replacing the case, I have officially Ship of Theseus’d my entire computer build, which started in 2009. Still running same Debian install that started in 2011. An extension of the self.
It would be an odd future if making robots with exactly the physical capabilities of humans ended up working best because that's what we have lots of video of to imitate.
It is hard these days to find an online sandbox where students can play around with a language model like GPT-2 or GPT-3, as they existed before instruction tuning.
The best one I can find is this HuggingFace space, but it's a suuuper small GPT-2 (137M parameters). Anything better?
I seem to remember you specifically talking about an impending AI reproducibility crisis as old models are phased out. Maybe a hallucination.
Things like this are why I push the old ways on my interns. If nothing else, it drives critical thinking about evaluation of prompt engineering tasks.
I'm so worried creators are all going to get punked by AI companies into stronger copyright to the point that death + 70 seems quaint. Let alone the consequences of overturning Author's Guild v. HathiTrust/Google...
I love working with students. Just the best afternoon. Always invigorating.
Born Digital Collections, Archives & Memory 2025 conference badge on a floppy disk
Brilliant floppy disk badges at the #BDCAM25 (Born-Digital Collections, Archives & Memory) Conference #DigitalPreservation #DigiPres
What did the United States look like in the 1970s? Why did a majority of Americans advocate for the EPA and a Republican-led administration support the agency? What did it mean to try to clean the air, water and land? Explore at digitaldocumerica.org
The 3-30-300 tree and urban space guideline
The short version of the 3-30-300 rule when it comes to Trees, and Green Infrastructure in Urban Areas
