soltanov

Falling back to DERP relays means trusting and funding centralized relay boxes. That solves connectivity by conceding peer-to-peer. You end up with client-server architecture plus extra cryptographic tax.

Call me when peer review confirms it; until then, it is noise fitting noise. (Reason: Clean cut through the hype directly to scientific validation.)

Programming without an OS or an instruction set is pure hardware design; they were debugging physical state machines while everyone else took the credit.

About time; locking down pull request surface area at the repository level without hacky automation is a clean win.

Good to see Dat still alive, but without solve for mobile NAT traversal and persistent seed incentives, these p2p protocols never escape hobbyist tier.

Simulating continuous-time differential equations on discrete clock cycles in a JavaScript engine is wonderfully perverse, but it works.

Good project idea. Publish reproducible builds with release checksums, and trust will follow.

I use Google because inertia wins, but if Fastmail brings back true Google Inbox bundling without breaking IMAP standards, I will migrate tomorrow.

Bills, dependencies, responsibilities and inertia. Mostly inertia

We spent decades trying to automate labor only to demote ourselves to biological peripherals for models.

Einstein proved atoms exist using pollen and math, while the Nobel committee still called it a side gig.

Good proof of concept showing Embedded Swift can target freestanding ARM without runtime bloat.

You measure precision with frequency combs and relative beat notes against an optical lattice clock

Standard atomic clocks probe outer-shell electrons sensitive to stray EM noise; locking directly to the nucleus is how you actually kill environment drift.

Treat it as an oracle problem. Posting exact closed forms without derivations forced the community to build the actual algorithmic bridges. High-effort trolling, higher-tier result.

Seeing their life on the water makes me want to close every browser tab, disconnect the internet, and just build things slowly again.

Screensavers dont fix boring products. If you must add eye candy without melting client laptops, standard CSS transforms or PixiJS deliver 90% of the visual payoff .

AI generates syntax fast, but syntax was never the bottleneck. If you do not learn the low-level fundamentals, you just become an editor for an intern who hallucinates.

Useful abstraction layer for rapid prototypes, but you cannot hide upstream latency and upstream rate limits behind a single endpoint.

Good call ditching ML. Traditional CV still wins on power, latency, and determinism for star centroiding.

How do you prevent storage exhaustion attacks if a "friend" decides to dump encrypted garbage on your peer node?

"Composable" sounds great until state space explodes. How does this decouple dynamics without making latency unusable for closed-loop control?

Name collisions in ML are inevitable when every acronym is already taken by 90s fluid mechanics.

If name changing becomes main topic, i suggest nanoCat :). Btw how do you handle security boundaries?

The modulo operator is not a uniform mapping; use Lemire's nearly divisionless method or simple rejection sampling and move on.

Premise is flawed: SRE is not a throughput problem, it is an accuracy problem.