Especially when the following years once again confirmed that none of the big enterprise customers care enough to cancel any contracts over QA problems.
A lot of the stability problems come from trying to scale a free product on a still WIP cloud solution without costing too much; the same software running on separate, paid for infra has a lot better odds.
IMO the constraint is that AI is still rather anaemic at sustainable green-field projects: It's good at one-off oneshots, and also at refactoring or fixing bugs or adding features to existing projects, where test suites and significant architectural scaffolding already exists, but the more you move away from that, the more wobbly the results get, and the more the human once again becomes the bottleneck, for all the hard work of coming up with all the conceptual scaffolding in the first place. Typing speed rarely was the bottleneck there anyway.
Reaching into ring -2 or the TPM allows privilege escalations past traditional "root permissions" and lets attackers defeat the sort of tamper protection that's designed to make escalations to local root manageable. Wipe-resistant malware, falsified cryptographic attestations, all sorts of fun.
I'm always flabbergasted when I see what people pay and how much effort they need to invest to keep their cloud websites from eating them alive.
My allegedly more complicated VPS stack needs an afternoon of attention every two years when a new Debian major release is necessary, and costs have been predictable for 15 years, no matter what happened traffic wise.
The post-release XP service packs, to be precise. XP's release was a perfect storm of events: Suddenly, millions of home users receiced an OS that had dozens of networked RPC services running in the background, and received DSL modems around the same time that exposed their computer directly to the internet. A slightly suboptimal combination at a time when software firewalls were a separate product you had to buy for a hundred dollars a year, if you even knew they were a thing.
Model quality is only one aspect, the bigger problem is making it work in a fully integrated enterprise platform, and Google has always been lacking when it came to the latter.
At this rate, if Google has a flagship model, you're better off plugging it into a competitor's tooling than hope Google figures out how to use it.
The primary advantage of dedicated hardware accelerators for the past ~15 years been power efficiency, rather than performance. Not much of a problem on a desktop, but very noticeable on mobile devices.
kubevirt is a cute joke, but hardly a replacement for ovirt or vmware. It's a real shame redhat keeps sabotaging itself to chase hypes with half-assed non-solutions.
"around" is the best way to describe it; the libvirt/virt-manager ecosystem isn't dead, but redhat killing off ovirt/rhev support drained a lot of resources out of it.
And for some bizarre reason people decided that the much less mature (both organizationally and technologically) proxmox VE is the best thing since sliced bread, so everyone who does care about linux virtualization is now trying to hammer some homelabbers' collection of perl scripts into a replacement.
Step one: Decision makers who can change these processes need to be aware of this problem. Many companies fail this simple task.
Step two: These decision makers must be held accountable for the success of the process. Many companies fail this simple task.
Step three: These decision makers must be willing to admit that they made a mistake, and risk loss of prestige and political capital. Guess how likely that is.
And the bigger the company, the worse it gets. It's a good thing we didn't go through 20 years of consolidations and mergers. Oh wait.
Yeah, Elon's whole shtick is make things scale up that everyone else thinks can't scale up. Sometimes it works (batteries), sometimes it works spectacularly well (Falcon 9), sometimes it fizzles out because it turns out everyone else was right (tunnel boring, solar tiles).
Jokes aside, I don't really see the appeal of a demake of helix; it only really makes sense to me as "vim but far more concepts integrated as first-class features rather than addons"; take that away and you have vi with less muscle memory. So, kakuone with the serial numbers filed off.
~90% of the failure modes grub can fix don't exist without grub. I can't remember any time when its needless complexity was actually a net benefit compared to literally any of its alternatives (and between gummiboot/syslinux/efilinux/isolinux/systemd-boot/efistub I've used a lot of them).
Grub is really impressive in how it consistently spent the last 30 years focused on improving everything except the UX of the one workflow 99.99% of its involuntary users need it for (boot linux as reliably as possible, and make it easy to debug when it does not).
Or use UKI and throw the current kernel to /efi/boot/bootx64.efi; there's plenty of solutions to sane bootloader/kernel management if you're willing to invest 15 minutes into the topic and not act like it's scary and complicated (it really is the opposite).
> The point was usually not usability. It was identity.
And we're not even getting usability out of it! Each of those bland react-angles is subtly inconsistent with the OS, with each other, and very often, itself. And in 6 months everything will move around again, for no reason other than to keep the responsible managers employed, without improving UX. And a11y is crying in a corner somewhere, forgotten.