rpcope1

That's bogus unless you're talking above 10G, in which case the only things typically doing wire level performance are expensive and sold by vendors like Cisco, Juniper or Arista (and maybe it's higher than 10G now). Frankly, the vast majority of ARM based routers, unless you go out of your way, are garbage. Any x86 built in the last 15+ years with a NIC that's also not trash _will_ push at least 1G routed with a non-trivial firewall on a modern Linux kernel no problem. I've still got half a dozen PC Engines APU2 boards that unequivocally will get right at 1G on Debian Trixie; newer ewaste like Dell Wyse thin clients or Optiplexes would absolutely smoke those boards on performance.

Setting up a router with Debian is really kind of overstated in terms of difficulty, and frankly half of everything wraps Debian anyways (looking at you Ubiquiti/Vyatta), so why learn someone else's half cooked configuration language when you can just do the real thing.

It's almost like TeslaCharts might have been on to something..

To someone not familiar with CXL that still gives the wrong impression. As far as I have seen, CXL is supposed to be cache coherent, and should require less invasive rework (if any at all) of applications to take advantage of it; that's part of the enablement of memory disaggregation that CXL is pushing towards (similar to the storage disaggregation push a decade or so ago).

While I agree the "Uncle Bob" cargo cult is IMO dumb, this really rubs me the wrong way:

You have the vendor pushed acronyms like: ACID, CAP, OLAP, OLTP.

ACID and the CAP theorem in particular has _very_ serious consequences if you're doing anything that isn't a toy with a database, and if there was one thing worth beating into developers, it would be the consequences of those. With the big push on columnar stores and various flexible database options (and replication and etc.) understanding that OLAP and OLTP workloads have very different needs and performance profiles is also something that many (most?) developers seem to miss, and again seems like something that needs to be beaten into a huge cohort of developers. Just because there are some silly/stupid cargo cult acronym around process, does not mean that anything that's an acronym or is confusing to your _garden variety history major_ can just be dismissed. This is part of why we get a lot of the monstrosities that exist out in the world today; there's this really perverse behavior I've seen among people coming from somewhere other than the hard sciences to wave away formalism and anything that smells remotely like math as "oh they're just being dogmatic".

PCB layout is an art, and doesn't seem to map well to LLMs (I tried for shits and giggles recently). Claude in general, kind of like code, does a lot of redundant belt and suspenders stuff in the schematics it generates (if it can generate them at all). It's one of those things that's really not there yet outside of the simplest designs.

You could start by pushing for RVs (both bus and gooseneck) to have to conform to the same standards as a commercial driver would. That seems like one of the most immediate and obvious places where it seems crazy (IMO) that we allow people to get behind of the wheel of stuff they have no business driving.

It's honestly frustrating that you can't buy anything like an S10, an O.G. Ranger, Jeep XJ, or even the 90's and early 00's GM and Ford full size trucks. I understand they were at least marginally less safe in a crash, but I think changes could have been made to get them safer and more efficient without compromising the ease of repair and generally more pleasant driving experience. It's amazing how much more divorced from the road and the outside world a new truck is compared to even the late 00's pre financial crash trucks, especially as any additional utility they provide is unchanged at best, and frankly so far as I can tell grossly net negative (i.e. how many 8 foot beds have you see recently? How many flatbeds have you seen that weren't on obvious fleet trucks? The Suburban is also frankly grossly less utilitarian now).

Ignoring the validity of the test, one of the more strange things I noticed is that apparently native English speakers only have a total vocabulary of 15k to 35k words? I probably live in a bubble, but that seems profoundly low.

Are we just talking shotguns or can it be anything they manufacture? Answer is probably Beretta though.

It's not just Windows where the rendering goes to shit immediate: any time I've got it open in tmux on Linux, it becomes a basket case in probably a few hours or less.

Are you conflating a router with SNAT? Routers as in L3 routing are not an "IPv4 only abstraction."

Having worked in places across both extremes (software engineer doing lots of other things including BD, hardware, ops, etc. to just being a JIRA ticket machine monkey), I am suspicious that HN readership is biased towards the former and frankly the bulk of "software engineers" in the world _willingly_ exist in the latter category. I didn't experience the latter until later in my career and God Almighty was it uncomfortable, but I think if AI were to displace some subset of "software engineers" it would those (they also seem to overwhelmingly dislike writing any prose whatsoever, which to me is a major tell). Many, many software engineers outside of hotshot shops seem either incapable or profoundly averse to "asking the questions" as you say.

I had the same thought, it's kind of like taking the guard off a 4 1/2" grinder. Real convenient until the cutting wheel explodes or the grinder gets hung and kicks back.

There's probably a lot to say about it's merits or problems, but given the demographics (or my perception of them) is largely "software people" can you really be that surprised or angry given that this could snuff out a _lot_ of people's livelihoods like nothing we've probably seen in our lifetimes?

Either get let go, fall behind, or pick up an expensive stimulant habit to try and eek out a little productivity at unholy hours.

I am maybe positing something even stronger: say you had two prompts, both with the same information, one was written in the style of a good paper out of Nature or Science, one written in the style of a bad Twitter post or other kind of mess, even with the same information, I increasingly believe even for the top end models like Opus, the results are at least materially different if not grossly so. I really believe the stronger English input yields demonstrably better outcomes, even if the information contained inside is the same.

That's been one of the gravest re-realizations I've noticed watching coworkers trying to pick up "agentic" coding: they often just break down into "just fix it" or "why is this broke". I've noticed that even though supposedly there's training or some sort of work done to make the agent work better with unclear or ambiguous grammar or bad structure, it feels like the quality changes palpably when you talk in clear well-structured English and provide at least a good background on the task. To me all of that feels natural, and I like writing and explaining anyways, but it's seemed like an almost insurmountable obstacle for some I've met (and I'm not even talking ESLs either). I strongly suspect those communication and writing skills will be a major factor in the bifurcation of haves and have nots as software "engineering" as we understand it continues to change.

Honestly, for certain classes of problems that have changed in the last couple of years, I've had good luck just finding decent academic lit that's shown up in places like ACM recently and feeding it in when working with an agent. Does it get everything right? No, but it gets you a lot closer and I've been pleasantly surprised how well it can integrate work that post-dates it's training if you finesse it a little.

Sure the people doing YouTube videos on fine carpentry (the ones that look like they just stole half of Woodcraft in some giant heist) are going to do everything themselves and do it what they consider right. It doesn't have to be like that. Ikea used to sell a pretty nice hardwood slab that I used for my brother's desk with a sturdy manual standing frame. I don't think we did much more than spray clear coat (and who knows he might have done something crazy like gel sealed it later). It was maybe an afternoon and a date with a hex driver and some high grit sandpaper, and it still looks better than what you can buy from Ergotron or whatever eurochic people are buying. Even common boards with a little bit of elbow grease and a few handtools can be made to look better than basically anything you can buy, and speaking from lots of experience you notice the problems for about a day before you move on (unless they're huge problems, which they rarely are) and think about something else.

All that to say, you might surprise yourself what you can do without a monster boomer wood workshop full of Festool and other unobtanium, and feel pretty good about it.

I mean if it has no moving parts, that's 100% true. Having spent too much time around the wealthy with "taste" I can't believe how much money people drop on dumb subpar shit when i.e. with the desk you could have spent a hundred or two for a high quality wood slab (or God forbid glued and planed your own), and afternoon with some good varnish and/or stain, and ordered or scavenged some nice commercial or educational table legs and both had something that looks better than basically anything else and can be actually customized to what you need.

I've got it running in an LXC container. Other than occasionally updating it, it's been entirely trouble free (I did need to work to get it out of the Docker container but that's a problem most won't have). Honestly one of the most useful and low trouble self-hosted apps I've used next to Dokuwiki. As far as hardening, I have not done a huge amount, but it lives on my LAN and is only reachable via VPN from the outside, which again works surprisingly well even with my Android phone.

I just take ZFS snapshots. I've restored a couple of times that way just to test DR and it worked pretty well.

Yeah the Solar City debacle was just one in a long line of crazy stunts that were pulled that if the SEC had any teeth at all, should have gotten someone more than a slap on the wrist.

Honestly, most of the time whatever the newest variant of the TI-30 is ends up being plenty (and what I have at my desk).

You say that like it's assumed that ubiquitous smart phones were obviously a good thing, when it sure seems like there's an increasing number of people questioning that assumption.

I actually noticed the same. Having it work on Mithril.js instead of React seems (I know it's all just kind of hearsay) to generate a lot cleaner code. Maybe it's just because I know and like Mithril better, but also is likely because of the project ethos and it's being used by people who really want to use Mithril in the wild. I've seen the same for other slightly more exotic stacks like bottle vs flask, and telling it to generate Scala or Erlang.

Mercedes feels premium and isn't full of bugs.

I see you've never actually owned or worked on a German car, especially in relation to even modest Japanese models. Maybe they were a little nicer inside in the 80s and maybe 90s, but "German car" and frankly "European make" is basically synonymous with "big expensive pile of shit that's an expensive pain in the ass when things start falling apart (which they seem to with increasing rapidity)." It's like the disease that plagued British cars for the longest time got contaminated with the German propensity to build overly complex monstrosities.

I mean if it were anyone else, yeah I might agree, but I think Salvatore is being genuine here (and have seen Claude do a similarly surprising job fixing ops issues).

You're absolutely right! But Brawndo has what plants crave!

Go to any mechanic thats been doing cars, especially if it's a focused subset of cars, for more than a few year and I guarantee they'll either have some modified HF sockets or wrenches or some home made tools. Probably don't want to cut up your SnapOff debt peonage tools, but a lot of the time they alone don't cut it.

Maybe we'll wake up and realize that putting WiFi and stupid "cloud enabled" Internet of Shit hardware into everything was an absolutely terrible idea.