j2kun

I'm mostly replying to your "standing on the shoulders of giants" part, when in some cases, the giant is the work you did in the past and forgot about :)

FWIW, you may enjoy https://www.argmin.net/p/lore-laundering-machines

I was reading another source that claimed this example was inspired by an existing (rational polynomial) example from the literature (created in 1999 by a Russian mathematician Vitushkin).

The seed is almost certainly Vitushkin's old rational "counterexample."

From https://claude.ai/share/22abed98-d9af-43c5-9881-b19e009a07b0

This is not quite lore laundering, but it seems to be close.

I think one counterpoint here is that you can still get value in using FHE in settings that don't involve scaling the number of users (e.g. B2B, public/private collab).

IMO a big unanswered question in FHE (which is my field so I care about the answer) is key management. If you have a service with lots of different users, and each user requires 100 MiB or more in evaluation key material, simply loading that key material from disk onto the GPU for each user's request adds a nontrivial overhead on the order of magnitude of the FHE computation itself. Not to mention the storage cost (for a 1B-user service, each requiring 1 GiB of evaluation key material, requires 1 EiB of storage, which is a hard sell).

So it seems kind of reasonable to me that if you want to showcase your fast inference for its own sake, you can sidestep the remaining engineering problems around key management by giving the user pre-computed key material corresponding to hot-loaded evaluation keys.

gaining any sort of insight

The server providing the FHE-based service does not gain any sort of insight. This is a key point: only the client can see the output of the computation.

There is also Michael Lewis's book "Who Is Government?" if you want more meaningful stories.

People do this all the time

https://www.reddit.com/r/dataisbeautiful/comments/1iu44ub/up...

https://usafacts.org/government-spending/

https://sankeymatic.com/gallery/federal-budget.html

Discovery is the entertainment for the rest of us.

FTA:

the number 12 features heavily in the Bible: Jacob’s 12 sons, the 12 tribes of Israel, the 12 apostles and the crown of 12 stars in the Book of Revelation are just a few examples.

And about "aspect ratios" (the article doesn't mention aspect ratios):

12 is a number particularly well-suited to establishing proportions, as it has many divisors

the apse, at 75m (7.5 × 10), followed by the vault of the transept at 60 metres (7.5 × 8). The nave vault is 45m high (7.5 × 6), the side aisle 30m (7.5 × 4), and the choir 15m (7.5 × 2).

neither 10 nor 8 are divisors of 12, and 10 has a factor of 5. TBH it seems like 15m is the base length here, not 7.5, then you get multiples 1, 2, 3, 4, and 5, but I'm guessing these sequential integers are not considered "beautiful" numbers by the author so they picked 7.5 to pigeonhole some divisors of 12 in there.

Call me grumpy, but numerology does not make architecture more beautiful.

In my view, LinkedIn has never really been a place for serious discussion.

I heard someone say recently that one reason tech sucks is because the sociopathic narcissists domesticated the hackers.

I think I agree, the FAR bigger impact on my book's sales was Google search deciding not to surface it in search results. Presence on pirate websites had no effect, and eventually I switched to the PDF as "pay what you want."

The main function of copybara is not moving code but modifying code to make it suitable for a different repository structure, build system, etc.

There are at least some technological solutions here, such as anonymous credentials. [1] Modern versions of this technique allow one to associate metadata (like a proof of age exceeding a threshold) in such a way that the verifier can't even correlate repeated requests across users.

Governments that are serious about age verification and individual privacy (which, doubtful they truly are) should agree on a protocol and set up certificate issuers that are associated with a digital ID. Then age verification will not be an invasive procedure or risk data leaks or insider threats.

[1] https://blog.cryptographyengineering.com/2026/03/02/anonymou...

For our situation, building a foundation of trust in our community is more important than attracting as many contributors as possible. If a one-time face-to-face introduction is infeasible, then there are many other projects to contribute to. (And this is considering that our community is all math PhDs, cryptographers, and compiler engineers; we are no strangers to neurodiversity.)

and if their bot is actually good? why not? fix

One reason: automating the construction of a "trustworthy" profile lowers the bar for attackers who want to plant xz-style backdoors. Not to mention polluting the various signals people use to evaluate candidates for jobs.

Indeed, a request for a short video call filters out most of the people who are looking to pad their resume with LLM-automated contributions, while adding an extra layer of welcome to genuine newbies who want to join the community.

In my main project we added a new requirement that all new contributors meet a maintainer in a non-textual format before their first PR is merged. Seems to work well for a small project.

Neat! I also enjoyed https://q.uiver.app/ by https://github.com/varkor which is a bit more specialized.

Yes, Steam Machine is optimized for gaming, but it's still your PC. Install your own apps, or even another operating system. Who are we to tell you how to use your computer?

Indeed, thankful for that.

Perhaps polyhedral optimization, tiling, scalar evolution, vectorization...

I guess garbage collection is pretty advanced in the syllabus.

Looks like there is quite a bit of overlap with regards to the optimizing parts between these two courses. I guess it's switching from the dragon book to academic papers that makes it advanced.

I'm a bit confused about what makes this course "advanced." Most of the topics (dead code elimination, data flow, dominator analysis, SSA form) seem like they belong in a first course on compilers.

What is kind of annoying is that the author jumps to "I hate compilers" instead of "I will report/help fix this bug upstream."

Instead, Google gave up on search engine development /s

Their whole argument is that AI's added efficiency means they don't need to set aside valuable human time anymore. Why can't they just point Claude at Claude Code and ask it to reduce memory usage by 90%?

We all saw their code...

I would say the same thing about stories purely about AI company stock price changes, but I imagine that is a losing battle here.