True car fans will appreciate the sarcasm. Fuck car week. 24hoursoflemons.com/concours-d-l...

Is peer review falling apart? (try to ignore the annoying animation at the top of the page) www.nature.com/immersive/d4...

‘A waste of time for all of us’: caught in the crossfire of the peer-review wars

Review mills — researchers who write fake referee reports with coercive citation requests — are setting off a war in academic publishing. What happens to those caught in the middle?

nature.com

Amusing tidbit about eprint.iacr.org: the query "guessing guessing" is leading a lot of people to eprint, probably because of eprint.iacr.org/2026/1377

Cryptology ePrint Archive

The Cryptology ePrint Archive provides rapid access to recent research in cryptology.

eprint.iacr.org

first thing I do in the morning when I look at my phone is check the news for mention of a big beautiful funeral.

I have managed to avoid ever flying through Dubai. I don't expect that to ever change.

I've never been bowling. Hacker News reminded me of this.

starting to get canteloupes in our garden

My license plate frame on one of my cars reads "I can't complain but sometimes I still do." My wife's says "We've been trying to reach you about your car's extended warranty."

People forget that album covers were works of art. So was the music.

I've written several LaTeX packages in the past, and also published a couple on CTAN. I just started a new one completely in claude code. If it can understand LaTeX expansion then I figure it can understand anything.

no you probably should not refer to this. orcid.org/0009-0006-64...

The IACR board has put forward a proposal to revamp the publishing of the general conference proceedings for Eurocrypt, Crypto, and Asiacrypt. iacr.org/hybridpropos... I would encourage members to engage in discussion about this on iacr.org/discuss

I have been using the paid version of claude for both code and mathematics research, and I'm very impressed by it. By contrast, I've been using the free version of Gemini and it continues to feed me nonsense maybe 1/4 of the time. I didn't have any expectation going into this.

Just a reminder that there's only about a week left to leave comments on the OMB proposal that will cripple federally funded research. Here's my attempt to convince you that you should, along with info on how to do so effectively.

Ars Technica@arstechnica.com · last mo.

@alaskaairlines.bsky.social has a completely useless website that is designed to make sure you cannot login. Luckily there are other airlines.

I know this sounds ridiculous to my east coast friends, but I'm visiting friends near the Monterey Bay area today. I may need to take a jacket. ❄️

In a continuation of my "Old guy yells at cloud" series, I note that bing thinks the title of eprint.iacr.org/2022/011 is too long for indexing. Kind of makes you wonder what the hell they are doing up there in Redmond.

Security Analysis of Coconut, an Attribute-Based Credential Scheme with Threshold Issuance

Coconut [NDSS 2019] is an attribute-based credential scheme with threshold issuance. We analyze its security properties. To this end, we define an ideal functionality for attribute-based access contro...

eprint.iacr.org

Dear 185.177.72.49. Why the fuck have you sent 2940 HTTP requests to eprint.iacr.org in the last two hours, all returning a 404? I'll give you credit for sending 34 requests that result in a 200 return code, but your hit rate doesn't seem particularly high. Go curl in a corner.

Cryptology ePrint Archive

The Cryptology ePrint Archive provides rapid access to recent research in cryptology.

eprint.iacr.org

a single IP address submitted over 20k searches to print.iacr.org yesterday. That's a WTF.

It has been an interesting experience to try and control crawling by AI companies the last few months. They are mostly well-behaved, but they seem to have recruited individual users to do their crawling for them. There are also a LOT of PhD students trying to find projects with public corpora.

There is something quite weird about Facebook telling me to wish a happy birthday to a dead colleague.

I have decided that Google #gemini is completely fucking useless for writing code. The amount of hallucination about code is a constant source of misdirection.

nesbitt.io/2026/06/26/i...

Incident Report: CVE-2026-LGTM

A series of unfortunate agents.

nesbitt.io

Call in the experts

I asked gemini to write a brief parody of the Citizens United decision by the US Supreme Court. It's not bad! sigcrap.org/2026/06/citi...

I've been using claude and gemini pretty regularly these days. My impression about using claude for mathematics is that it spews a lot of nonsense, and has a strong tendency to want to confirm what the user said. I am used to having my collaborators object when I make a stupid suggestion.

I've been trying claude code for writing code. I find that it's sometimes slower than what I do, and it creates a lot of errors along the way so it takes a lot of effort to fix things.

I have an old guy collaborator who cannot fathom how to do LaTeX on his own machine. He has been mooching off the paid overleaf accounts of his coauthors. I could run my own VM for overleaf, but the code is so gigantic that it would cost as much as a paid overleaf account. I think I'll try prism