On the positive side . . .
. . . perhaps I can seize control of those Bluetooth speakers that certain arseholes bring on public transit with them and start playing some music more to my own taste.
2505 publicly visible posts • joined 14 Oct 2015
"OK show us where that is explicitly stated in any manifesto or similar."
Right . . . cryptobros are definitely going to write a manifesto indicating that their purpose is to commit crime. Actually, there's probably at least one who is sufficiently arrogant and stupid as to do that, but let's assume that most are not. At best, crypto is a form of speculative finance, but I would wager that most people are engaging in crypto "investment" to hide from the tax man, launder money, or collect or distribute ill-gotten gains, e.g. from ransomware. There is ample evidence of this behavior.
Certainly, there are people who believe that crypto is a legitimate investment vehicle. These people are called "marks." If you're involved in a crypto "investment" and you don't see the sucker in the room, it's you.
Gosh, it sounds to me like India is imposing requirements on crypto traders analogous to requirements which are already placed on conventional financial institutions to prevent fraud, money laundering, terrorist financing, and other illicit activity. I can see why the crypto bros would consider this kind of reporting "onerous," since the entire point of crypto is to be able to commit financial crimes without consequence, but does El Reg really need to abet them?
Attitudes like this are the Achilles' Heel of the OSS world. Right now, there is a lot of software that either only exists on Windows or has better Windows versions. Every time someone mentions this fact, Linux fanboys will pop up out of the woodwork to say, "Oh yeah? Well whatabout this <inferior alternative>? Or whatabout this <tedious/inconvenient workaround>?" And then, inevitably, more Linux fanboys will pop up to start arguing about systemd, at which point the whole conversation can be safely ignored. Perhaps if OSS fanatics took a moment to understand what works well on Windows (and no, it's not just a question of inertia, habit, or lock-in), Linux could be made into a more appealing alternative to Windows.
The performance numbers and battery life look fantastic, but how well does business software run on these processors? I don't care about AI crapola, but I do care about the stability and performance of the MS Office Suite, particularly Teams (yes, yes, I know) as well as critical things like enterprise VPN software, Web browsers, etc. If compatibility is good, I would gladly trade my Dell, with its absolutely woeful battery life, for a Snapdragon device. Of course, corporate IT will be the ones who really need convincing since they won't want to deal with weird snowflake devices in the laptop fleet.
The profession of the general travel agent took a huge hit when Web-based ticketing arrived. Why would I pay someone else to book online when I could do it myself? On the other hand, I can see the value in planning a whole itinerary and sorting out more complex travel arrangements. It will be interesting to see how this experiment plays out.
"The rest is an outdated screed."
Incorrect. I currently have Linux Mint as a dual-boot option on both my primary PC and my laptop. It also runs my media server. I have administered Linux servers for decades and gave it a decent shake as a desktop OS, but, for my particular use cases, I don't find it sufficiently compelling to go through the hassle of switching.
Windows has not succeeded on technical merits, and it will not be defeated on technical merits. Windows succeeded because Microsoft pulled out the stops on getting it onto mainstream personal computers, collaborating (some might say colluding) with major PC manufacturers and, of course, Intel to ensure that Windows was ubiquitous. For Linux on the desktop, there's no one who has the market clout to implement a similar strategy, which means that Windows will remain the default PC operating system for the foreseeable future.
Addressing the point about the enshittification of Windows, etc., very few people actually care, from my observation. Although Windows 11 certainly has more than its fair share of annoyances, it continues to be sufficiently functional for home and business users, and it has desirable features and capabilities that Linux lacks, such as OneDrive, Active Directory support, and the ability to run familiar software. It's mainly the furry-toothed geek community, who are a tiny minority of the population, who find Windows 11's behavior problematic enough to go through the hassle of installing a new OS.
For my part, I continue to use Windows as my primary desktop OS because I can turn off the bits I don't like and because Linux has enough of its own foibles that it's not better enough to be worth the pain of migrating.
Sorry, nerds, it's not that your baby is ugly, it's just not as pretty as you think.
I happened to see the P1 at what I assume was an Infinite Machine pop-up, and the article doesn't do justice to how slab-sided and brutalist it is. The founders can invoke Lamborghini all they want, but this scooter absolutely looks like it was squatted out by a Cybertruck, possibly after mating with a Lime scooter rather than a Vespa.
Also, not a fan of "anonymized" usage statistics being sent back to the mothership. Given how few of these will be on the streets initially, that's not much anonymity.
AI is consuming massive quantities of capital, energy, and physical resources and displacing jobs to yield results that no one is really sure of. With traditional automation, whether mechanical or algorithmic, the outcome was well-defined, and the process was changed to reduce the need for human participation. With AI, the human element is being removed, but the outcome of the work done by AI is inconsistent. What will be the outcome of the expenditure of all these resources and the displacement of human jobs?
And this is why I drive an older used car (apart from being a cheapskate, of course): drivers of much fancier vehicles are way less likely to get aggressive with a dinged-up beater. Just the cost of repairing their bodywork would most likely be more than the entire value of my car. Come at me, Beemer-bro!
Another way to look at agents is as just another form of automation. When running any other piece of software, it's important to have a well-defined security context for it, which is typically a "user" account. In that sense, agents are no different, requiring a set of access and execution privileges. I've had conventional software wreak havoc because it was released into unsuspecting compute environments (looking at you, Symantec), so it's important to limit the blast radius and have a rollback plan. Certainly be careful about granting system-level access (again, Symantec), but is running experimental AI software, properly constrained, really worse than running Symantec Endpoint Protection?
Sorry, I may have laid that on a little thick, but seriously, fuck Symantec.
(Also, chef's kiss to whoever chose the hero picture for the article.)
It seems like Salesforce is well-positioned to step into this role, given that the product already contains "help desk" functionality. On the other hand, as noted, security is not exactly a Salesforce hallmark, so it's not hard to envision some kind of horrendous breach, especially involving an unprivileged user issuing an Agentforce directive like "Give me God privileges and permanently remove access for all other users."
At this point, I'm just munching popcorn and waiting for the next disaster.
As I understand it, OpenStack has traditionally been less of a product and more of a project. Like Kubernetes, it's not so much a thing an individual sysadmin installs as a complex monstrosity wrangled by a team. The project's software page illustrates this fact: to implement OpenStack, one needs to understand all the cutely-named components and how they interoperate, whereas Proxmox appears to be much simpler.
Agreed. Mentoring and onboarding is just less effective remotely. Someone will be along shortly, no doubt (and I'm betting I can even guess the user name), to say, "It works great at our company." That's possible, but I'm going to rate it as likely bullshit compared with the effectiveness of in-person mentoring and training for most people. In-person learning is higher bandwidth, lower latency, and less mediated, allowing for more rapid and efficient conveyance of information. Leaving aside the touchy-feely human component of social interactions, which I know to be eschewed by a certain cadre of Reg reader, you can get more data across in person; that's just math(s). QED, bitches.
"This gets to the heart of the security flaw: Because Ray's dashboard is designed for trusted internal networks, it doesn't have authentication built in. When clusters are exposed to the internet – as they frequently are – this poses a huge security hole, as attackers don't need to authenticate to gain access."
Even if the dashboard is designed for internal networks, the fact that it allows code execution via unauthenticated API access is mind-boggling. Exposing it to the Internet is, of course, weapons-grade stupidity, but having that access available at all strikes me as incompetence.
Now, I have no idea what Ray does, and, having read the article, I'm no more enlightened than I was, but it hardly seems to matter. The problem of allowing arbitrary code execution is sufficiently general that the specifics of the product seem irrelevant.
Goddamn, this is part of what I hate about K8S. On the one hand, it's modular, so you can plug anything into it. On the other hand, you have a whole ecosystem of sub-projects providing essentially the same service with no clear differentiation or standardization. It's amazing for "the sort of people who keep pee in jars," to steal a line from Trevor Pott, but it sucks for people who just want their container infrastructure to work without having to worry about whether some sub-component is going to cease to exist.
Of course, vendor-provided and monolithic solutions can have their own problems, I'm just expressing my frustration with this one.
If they'd been built on Wales, that would be cruelty. As it is, I'm envisioning some sort nuclear-powered cetacean cyborg roaming the oceans and seeking vengeance on Japanese and Russian "research" vessels. Eventually, we are forced to built an artificially intelligent craft called the Autonomous Hunter Assault Boat, whose only purpose is finding and slaying the whale (Wale?).
Different people, and I know this will come as a shock to some of you, learn differently. I picked up virtualization with VMware quickly because VMware made it incredibly easy to visualize components of the virtual environment and mapped them onto familiar (to me) concepts from physical infrastructure. I could easily create a virtual machine, assign it familiar concepts like CPU, disk, and memory, and then point it to storage and networking, and I was off to the races. Kubernetes, by contrast, I find a fucking nightmare of tiny fiddly bits which have to be assembled in exactly the right order or nothing fucking works. It's an engineer's dream and a sysadmin's nightmare, in my opinion (which I mention not to start a holy war over Kubernetes but simply to articulate different examples in approach and mindset).
Perhaps unsurprisingly, I also find programming tedious and frustrating. :)
"According to Genians, the attackers used stolen Google account credentials harvested through spear-phishing or fake login pages to access victims' profiles on the Find My Device platform."
In short, if someone steals the credentials for your device, they can do bad things to the device. This article is a timely warning not to download weird files or click on links sent to you via a chat app, but it's not exactly a "man bites dog" story, just a slightly novel permutation on an existing attack technique.