non-disparagement clauses doing a lot of heavy lifting here

Cool idea, especially the plugin security angle, but WordPress’s real strength is its ecosystem. That’s going to be hard to replicate.

RPKI makes BGP safer, not safe. It helps prevent some hijacks, but attackers can still mess with routing paths. Feels like we’re patching a trust-based system rather than fixing it.

“More neurons = intelligence” always felt like an oversimplification. If that were true, we wouldn’t be surprised by birds or octopuses anymore.

It's not a 1:1 relationship but they are related.

“Just use OPNsense” is great advice for production, but terrible advice for learning. This article is valuable precisely because it shows how little magic is actually involved in routing.

I would recommend VyOS Stream for this situation. It has better performance and hardware compatibility than *BSD-based software routers, and it also has a nice CLI that is syntactically similar to Vyatta and EdgeOS (found on Ubiquiti's Edgerouter line).

In additon, compared to PF/OPNsense or OpenWRT (Linux based), you have more control and exposure to the underlying network concepts with VyOS. You're not configuring the kernel manually, but you still learn quite a bit.

SOHO toys don't do routing in a real sense at all

IPv6 isn’t that compelling on a simple home network, but avoiding NAT and easier end-to-end connectivity are pretty real advantages.

Petitions don’t do much on their own, but they’re often how pressure starts. And ‘not European issues’ feels off when these companies operate globally anyway.

Pressure is building, thankfully. It's not just petitions now, but legal groups getting involved, etc. At least in the UK. Hopefully it spreads like wildfire around Europe. The orange Oompa Loompa is likely helping kindle those flames nicely.

The pill tabs are what get me too. I can ignore most visual changes after a while, but those somehow manage to feel both more distracting and less informative at the same time.

I figured they would do this when SoftBank acquired them.

SB likes 10x buys.

This feels less like automated research and more like structured trial and error with a decent feedback loop. Still useful, but I think the real bottleneck is how good your eval metric is. If that’s weak, the whole loop just optimizes for the wrong thing faster.

Designing a good fitness function, a tale as old as time...

I mean, isn’t that “the scientific method”?

Partially—but science also questions the hypothesis and the metric. This mostly assumes both are correct and just optimizes within that box.

Only if the model is actually a human or equivalent, otherwise we don’t know what it is.