AI In

This person is viewing the outcome discussed as a good thing, and I do not agree with that. Their politics are not mine. However, I do concur that this will be the result of AI infiltrating the economy.

And it is going to cause a lot of economic devastation.

Prompt Gen

Indeed. AI improves every day. It actually shocked me how much better it’s gotten, and how quickly. The first time it generated a fully-working Firefox extension that had never existed before with only one prompt, I was like, “Dang, I was wayyyyy wrong here.”

T Tao

If the best living mathematician says it, all y’all dipshits, mooks, screechers and losers should listen and believe.

AI is only getting better, too. We’re at the beginning of the beginning now. Always wise to remember that.

Junior Dev

This is because AI is currently “only” as good as a very good, tireless, junior developer. I can get someone/something like that to do whatever I want because I have an extensive technical background and have done near-dev work my entire life starting from when I was five years old.

Someone without that experience like Katie is not qualified to tell a junior dev (the AI) what to do or how to do it, so their efforts will fail. AI and junior devs currently share a lot. For instance:

1) A junior dev/AI has no idea why it’s doing anything, even if you explain it it. Can write good code but has no design sense or architectural vision.

2) A junior dev/AI will make whatever changes they are told, but will not be able to contextualize them in any real way.

3) A junior dev/AI will not set up its dev environment competently without extensive assistance.

4) A junior dev/AI will not understand what to test or why, especially in the context of the larger user experience.

5) A junior dev/AI will have no idea about security, and not understand it it any real way even with extensive explanation and guidance.

I am not saying that AI is not useful. What I am saying is that AI is only as good as a talented but green junior developer with 1-2 years of experience. That’s still far, far superior to nothing.

So that means that someone like Katie is not yet able to steer it correctly to build anything good or useful.

Fworking

Damn, is that the best they could find? Approximately 90% of the my company’s new product was written with AI. That’s of 120,000 lines of code. It’s not a simple product and not anything you could create by one person goofing around casually with Claude Code. It took a whole team of 8ish developers nearly a year. I’d estimate without Claude, the same project would’ve taken 2-3 years.

AI is changing everything and some people are just too clownish to see it. This is not said as an AI supporter or detractor; it’s said as someone with a fucking functioning brain.

Want To Admit

AI reaches the threshold of average human creativity.

That’s a pretty low bar, to be fair. The average human is nothing special in this or any other area. But still it is quite an achievement and AI will only improve.

The best AIs now can surpass 95%+ of the juniors I’ve ever worked with, and about half of the mid-tier folks. That’s substantial and is going to change the job market and society more than most people now realize or want to admit.

The left’s inane screeching about it won’t change that, nor will the right’s attempt to strangle the future in other ways.

Low Hire

Unlike the dipshit in the post below, someone who actually knows what she’s talking about.

Best I can figure, consolidating BLS/FRED sources, the US tech industry has lost around 700,000 jobs since the peak in late 2022. And that is a huge amount.

NAT Not

IPv6 is not insecure because it lacks a NAT.

Not this asinine shit again. I hate this idiot and idiots like this in general. That is, the “Well, ackshually” shitheels who ignore how anything is in the real world, standard practices, and how things actually work. And also do not really understand the tech, either.

First of all, you stupid motherfucker, a device can (and most consumer crap does) implement NAPT/PAT with dynamic state but often has1 no explicit packet-filter policy engine (what most people would term a “firewall”), yet will still refuse unsolicited inbound flows simply because these flows donโ€™t match any mapping/state. That is in fact de facto protection via reachability restriction. And that behavior is explicitly defined in NAT RFCs. The NAT RFCs in fact directly discuss filtering behavior associated with NAT operations (not just a separate firewall). Check out RFC 4787 (BCP 127), RFC 5382 (BCP 142), RFC 5508 (BCP 148) and RFC 7857 for how NAT really works. I’ve read those documents in toto several times over the years. I can guarantee that doofus has not.

Miraculously, he is right that โ€œNAT isnโ€™t designed as security,โ€ but the clown-ass shitstain then uses that to imply โ€œNAT adds no security value,โ€ which is false in actual practice. Nearly every existing IPv4 NAT (NAPT/PAT) gateway2 enforces stateful inbound blocking out of the box. This NAT — independent of the router’s firewall function — does provide decent default-on security for home users.

On the other hand, his core premise (โ€œmodern routers default-deny inbound IPv6 anywayโ€) is absolutely not a sure thing. Standards and real deployments often have non-optimal defaults, including configs that default-forward unsolicited inbound IPv6 traffic. This is because unlike IPv4, IPv6 expects end-to-end connectivity. So that means many router vendors ship equipment that way. Thus, having NAT adds quite hardy extra protection in practice. That is to say, with any IPv4 home NAT you need both a firewall hole and a port-forward/mapping mistake to expose a device. With IPv6 global addressing, exposure can occur with only one minor screw-up. Then boom, your whole network is out there on the wide-open internet.

This disphit’s NAT explanation is also crazy sloppy (he frames it as mainly destination-rewrite based on static port forwards), just glossing over or ignoring that the real โ€œdefault denyโ€ effect largely comes from dynamically created state. He overstates a conditional truth (โ€œIPv6 is fine if you keep equivalent edge filteringโ€) into an unsupported and often-wrong universal claim, using cherry-picked vendor defaults as if they were always the case. Also, he deliberately handwaves away as irrelevant the safety margin NAT provides in reality every damn day.

NAT wasn’t designed for security, wah wah. Carbon steel wasn’t designed for armor, either, but we use it for that in the real world.

My conclusion: Fuck this fucking clown who doesn’t know a damn thing, and what he thinks he knows is wrong. Read the RFCs, motherfucker. I’ll wait. You won’t understand them anyway, but I’ll still wait.

  1. And does not require.
  2. I have not seen one in 20+ years that does not.

Au Hell Yeah

I guess I am a tech elder now, and the reason for gold is that it was corrosion-proof and is easier to solder and wire-bond (the gold was present in more than just the cap). The older process (prior to gold) with tin and whatever else, was prone to oxidizing and corroding.

Gold was the best option at the time vis-ร -vis the drawbacks of the others and technological capabilities. Not needed any longer as the tech matured. I know this stuff because I read about it on some website back in like 1997 or so (maybe HardOCP).

Dev HD

What Twenty Years of DevOps Has Failed to Do.

One of the primary reasons DevOps failed is that it invariably ends up becoming the developer help desk. Into that department is dumped all the tasks and project that developers don’t want to do, don’t understand how to do, or are not smart enough to undertake.

And in my experience devs are even worse to support than regular users. Most non-dev users are aware of how little they know. However, developers often think they know much more than they actually do. This leads to untold problems.

DevOps is a failed experiment. In the end, ops people should do ops and devs should do dev work. AI might change that some, but that’s the basic shape of it.