Posts by doublelayer

Computing's big question for 2023: How many more questions can we endure?

Re: With every Amazon order:

While we're complaining about Amazon, I'll join by complaining about how they've mixed in refurbished items with search results. No, I do not intend to buy refurbished hard drives. Other things, maybe, but not hard drives. Disabling this takes several clicks and will be switched back with the next search, and you will be doing multiple searches because Amazon's search system is terrible, relying on the searcher to filter out obvious crap. If I keep finding new things to complain about, it will get even more boring, so I'll stop with that one.

Why would a keyboard pack a GPU and run Unreal Engine? To show animations beneath the clear keys, natch

Re: Useless feature

A wireless keyboard emits a signal that can be received at some distance. A wire emits some signal based on signals moving through it, but is not intended as a transmitter and is relatively hard to listen to. If you're worried about it, you have a lot of other wires where more important signals are sent, such as all those ones inside your computer that will carry the information about what keys you pressed after they were decrypted as well as other data. That is easier to hear than the wired keyboard.

As for what is easier, encrypting or shielding, shielding is easier. You can make shielded cables and many have. It increases component costs a bit, but that's basically it. If you make an encrypted keyboard, you need a local driver that decrypts it. You need a method for making sure the driver and keyboard are in sync (they can't just share static keys because it exposes you to repeated text attacks), and that needs to be ported to any OS you might want to use a keyboard with. If you try to use it with BIOS or firmware, that's definitely not going to recognize it. If you have an encrypted keyboard, it's going to need a "fall back to standard keyboard" switch, and everyone is going to leave that switch in that setting.

Re: Useless feature

It's most definitely a useless feature. Your suggestions, also, are not so useful:

"improving firmware security (Rust firmware": Why? Rust's security benefit is avoiding memory safety problems, which do indeed lead to security problems like buffer overflow vulnerabilities. That's a lot more important when input is received from another program, but you don't tend to get multiple programs running on a keyboard. The method for attacking it is mostly absent, and would either rely on any programming interface (which may not exist and is likely not prone to buffer overflow anyway) and user key presses (requires physical access). Writing keyboard firmware in Rust isn't likely to do anything different.

"encrypted USB connections": Not why, this time. More what? I mean you could set up an encrypted serial connection to a program on the computer which decrypts the signal and simulates the key presses, but why do you want to? The only way sending an unencrypted signal over a cable is a problem is if there's listening equipment quite close to the cable. That equipment could also use acoustic methods to hear what keys you pressed if it can't point a camera at you. It's not like wireless keyboards that need encryption both to avoid a more passive listener and collisions with other keyboards. Encrypting your communication on a cable that's right next to you just makes the keyboard harder to use with a computer without doing much for your security.

We don't need that much security on keyboards. Focus on the stuff they're connected to.

Re: Hackable?

I have the same question, but on the opposite side. How easily can I hack it? Not just installing a new theme, but doing some computing of my own on it. If it's got a CPU and GPU in there, that's kind of pointless unless I can compile some code without having to limit myself to their theming system. This keyboard isn't for me, but if I was going to buy it, I'd want to be able to write firmware which takes control of all the hardware and lets me do arbitrary things with it, not just changing the lights.

Too big to live, too loved to die: Big Tech's billion dollar curse of the free

Re: A couple of solutions, offered for free.

"and the only difference between running a desktop 24x7 instead of running a server 24x7 is......"

The difference is that "desktop" implies a machine you're using as your personal computer, doing other things, and server implies something you've set up for this task and possibly a few others, with the goal of a more stable system. The shape of box the hardware's in doesn't matter too much. In my summary of the tradeoffs, I specifically mentioned that a desktop is going to be inside your home network which is probably less reliable than you need and that you may be doing other things with it that impede the system's uptime.

Re: A couple of solutions, offered for free.

In order to do it, you will need:

1. A desktop that you intend to run all the time, taking on that power cost.

2. A desktop that you don't expect ever to have problems that would interrupt your service, meaning that using it as your own desktop where you might damage something or even need to reboot isn't a great idea.

3. A network connection that stays on at all times, including when you're not there.

4. Your ISP to give you a dedicated address which you can host stuff on on your residential contract.

5. Some equipment such as a UPS to automatically recover from power failures.

6. A plan for what you'll do if your power or network goes down when you're not there.

7. The technical ability to run the mailserver software.

A lot of people are missing one or more of those. My ISP doesn't let me run whatever servers I want. I don't have a plan for dealing with a downed network connection if I'm traveling, but I'm also not willing to let my email become unavailable until I come back. I could manage the rest of it easily enough, but these problems cause me to use a server located somewhere other than my house. The general public is likely to lack even more of these items.

Re: Hope

"It's the same way that I'll see stuff being offered for free on Craigslist and it's obvious that the person is really looking for somebody to come by and haul everything away at no cost to the person posting the listing."

That's sort of the point. You can have this stuff that I don't want, and I don't need you to pay for it, but you need to come get it. You are incurring a cost in time to obtain a useful item. I've had people retrieve unwanted equipment to scrap it for parts, and it is most definitely their responsibility to deal with the bits they don't want. If they don't want the item, they won't go and get it. This is nice because I don't have to throw things away when someone will actually use them.

Re: …puzzled?

Not all big religions have confession systems, and those that did didn't have organized systems for sending important confessions to someone who could handle them. Not that it wasn't ever used for blackmail purposes, but that's not what made religion powerful.

What made religion powerful was that people believed in the religious figures and trusted the clerics knew what those figures wanted. If the clerics say that the guy who can sentence you to an eternity of torture wants you to donate a lot of money, and you believe that the torture is a real option and that the cleric is speaking truthfully, you give a lot of money. Bring in social and legal pressure for people who don't believe one or both of those things and you have pretty good coverage. You don't need to leverage confessions for that, especially as people who don't believe in the power of the religion are unlikely to confess something they don't want someone to know.

That would probably work, as long as the ISP resists the temptation to make it a significantly larger bill or charge per month. I've never used an ISP-provided email (I don't think my current ISP even offers them), but I see enough of the things around that people must lose stuff when they let them expire.

Re: A couple of solutions, offered for free.

It does, but I had the misfortune to use that and won't again. I had a PDA in the mid-2000s, the kind that had WiFi (802.11B) but no cellular connectivity. It had a mail client that only supported POP. Therefore, my choices were as follows:

1. Download all the mail currently there to the PDA, with delete from server enabled, then handle it all there. Using a computer later was a pain.

2. Download all the mail to the PDA with delete from server disabled, and then have to deal with it all again on the computer when I used that. Nothing synced information about which messages I'd deleted, moved, responded to, etc. I used it, but as soon as an alternative became available, I stopped.

Re: Centralised service, centralised problem.

"Expect to see a decoupling of the 'Google account' used for sign-in to phones, apps etc. from the actual email service well ahead of this so as to limit potential regulatory issues as much as possible."

It basically is already. You can set up a Google account with a different email address. It's just that almost all Google accounts are Gmail addresses and that, if you go through the typical registration process, they ask you to choose your Gmail address instead of entering your own email. You can still get to that form if you want, and if it became a regulatory issue, they'd start saying that more loudly.

Re: I wouldn’t know how to pick one that was likely to stick around.

The number of complaints about a service is also correlated with the number of customers. If you find a service with few or no complaints, it might also be so small that losing a few customers could take it down. If you can find one with a lot of praise and few or no complaints, there's a reasonable chance they made up the praise and also have few customers. There is no simple solution for figuring out how long something will stay up. You have to research them more thoroughly and have a plan for what you'll do if they tell you they're going away (and hope that they do tell you).

Re: A couple of solutions, offered for free.

That would be nice, but let's be honest, it wouldn't do much for the e-waste problem. I'd be able to get several for next to free as everyone else abandoned them. Most users don't want to set up a self-hosted server for it, so many would be junked anyway. Also, if Amazon just gave out the keys and let people play around, even fewer people would actually do anything, as they'd have to work out the way to best make use of the hardware. Properly open sourcing the platform, including nicely documented drivers, examples, and backends, would take a lot more effort that it's clear Amazon won't bother with.

I like it when companies release source for stuff, but few of those projects ever result in a useful result. Reverse-engineering replacement firmware is hard, and there aren't enough skilled and inclined people for each one.

Re: A couple of solutions, offered for free.

Feel free to use only one, but don't expect others to do so. Being able to read my email on a phone when I'm out is useful. Being able to read it later on a laptop when it's more important is also useful. I'll continue to use multiple ones, but you aren't required to do that.

Re: A couple of solutions, offered for free.

The critique wasn't about whether to use Google, but about whether self-stored mail data was going to work for the general user. I run my own mailserver, and it's a server. It's online at all times to receive data, stores it there, and I have to pay for it. Running that off a desktop isn't feasible, and a lot of people don't have the expertise to do that. Whether you use Gmail or not, it's likely to keep using servers rather than self-hosted on clients.

The era of cloud colonialism has begun

Re: VC

I'm not sure if that's a requirement from all of them, but there is a potential reason for VCs to ask about cloud hosting, namely scaling. VCs are frequently obsessed (sometimes correctly and frequently not) about getting this business to grow really fast. Some businesses can scale just by adding a few more servers, but if it's something that requires a lot of capacity in local areas, then expanding worldwide can't be done so easily if you self-host. The business would have to hire people to either set up colos or datacenters in multiple places, including hiring maintenance staff and therefore working with labor laws in all those places. A larger company can manage that, but one that's just starting out will find it a bit more difficult. VCs may want some indication that there is a plan for adding geographical regions without incurring that cost, and proving that cloud can be used, even if the first regions are self-hosted, is a way of indicating that. Just as their reasons for wanting growth they won't get can be wrong, so can their insistence on cloud, but it does make sense at times.

Don't lock the datacenter door, said the boss. The builders need access and what could possibly go wrong?

Re: rebooting the system

The other thing about that particular exchange is that the laptop concerned was connected to the office's WiFi network anyway, so pulling that cable would disconnect the screens and keyboard connected to it, but would not take the laptop offline because it would switch from a wired connection to real WiFi. I don't know whether they knew any of those facts and whether they even expected that disconnecting the cable would have an effect on their internet connection.

Re: rebooting the system

Yes, this is the other side of the coin. I've had occasions where I've said something along the lines of "If you don't start reading that word for word, I'm going to make you spell every word in the message". That usually gets them to read exactly what it says, although in a particularly angry tone. Of course, by the point I have to say that, I'm annoyed even if my tone doesn't reflect that, and on every occasion when that's happened, the error message contains important details. There will come a time where the error message is one of those vague ones and I've needlessly accused the user of summarizing it unnecessarily, but that hasn't happened yet so I'm happy to apologize when it eventually comes up.

rebooting the system

I would say we should do something about people who don't know what technical terms mean but try to use them anyway, but only the BOFH's methods would work at this point and most of them are illegal. Just as this accountant thought the emergency shutdown was a reboot the system button, I've had far too many discussions with nontechnical friends asking for help who string together terms in a way that means nothing.

The most recent one I remember is "disable the WiFi system", which could mean about five different things but could not mean what they wanted to say, which was "disconnect this USB-C cable which connects to a dock which has an ethernet cable connected to it". I have also recently informed someone that their "tape backup system" uses disks and isn't turned on and that "the email server was erased" is not an acceptable phrasing of "I can't find the message I want in my trash folder because I've never used the advanced search feature". These are just recent examples. I've heard stupider and I'm guessing those reading this have heard much worse, especially those who work in IT (I'm a software developer, programmer, or software engineer, but none of these terms is properly understood by my acquaintances either).

SEC: Startup had 'no functional streaming service', raised $1.3m anyway

Re: How far did they get

This case is fraud as are most of the ones I described. There are some startups that manage to be so clueless that they don't get anything, but that's more in the realm of negligence than fraud. The key is that the operators of the company committing fraud know where the money went and the negligent idiots probably don't have any idea. Either way, there's likely to be lawsuits.

How far did they get

I'm wondering whether they just took the money and ran, always a popular choice. There's the other one I've seen happen: some non-tech people have an idea, start defrauding investors, then eventually hire one or two technical people (either fresh graduates straight off their degrees or my mate who fixed my laptop once) and tell them to build the entire service because "How long could it take to build a thing like Netflix?".

I'm never sure whether the companies who do that are just really stupid about how any project is built (I've seen people not trying to defraud investors who make that mistake, so it's at least possible), or are attempting to have some scapegoats to explain why their service didn't get built.

Pine64 takes another shot at an open tablet after chip shortages killed first PineTab

Re: longevity

"they're not going to be a goer, as far as I'm concerned, until they're running a long term support Linux OS."

That's sort of the point, in that they're designed to be integrated into the mainline kernel. That means that you can install any version of that (until thirty years from now when the architecture is removed for obsolescence). If good software is made for the device, patching it and keeping it maintained should be significantly easier than any Android device has been.

The real problem is in getting the support in the first place. We don't really have any polished mobile Linuxes. Right now, you can probably get any buggy missing-lots-of-features distro and keep it maintained with kernel patches for a long time, but it doesn't have the smoothness that Linux on the desktop has. I'm not sure whether this will ever be fixed, but if it is, it will require hardware to run the beta versions on so that developers aren't limited to Android phones with custom kernels. I want that to happen, so as many reservations as I have with Pine64, I still have to support them as they've done a lot more than the only other player has and I don't see any other companies stepping up to improve on it.

Don’t expect a Raspberry Pi 5 in 2023, says Raspboss Eben Upton

Re: Cheap media player...

At some point, you have to consider commercial media players. I went to some effort to get a Pi to work as a media player for a family member, getting the DRM support in, having a local server for their video files, and a nice tiny USB keyboard as a remote control, only to find that they weren't enjoying using it. They had replaced it with a basic Android TV stick which cost less, had a faster processor, native support for DRM, a client app that could talk to the local server, and came with a remote control they found more intuitive. I couldn't find any argument for why the Pi was preferable, and it meant I got another Pi.

At some point, building everything myself leads to poorer products than getting one that was built for the purpose. They're still brilliant for something that doesn't exist commercially, but if I ever need a media player, maybe I will consider not DIYing one. I've reinvented a lot of wheels, and they don't always provide a benefit.

Re: Bye bye Pi

They are comparable, but depending on the use case, there are some major differences. Both support WiFi and technically there is hardware support for Bluetooth LE on both, but the ESP32's Bluetooth stack is thoroughly tested and the Pico's is absent (hardware support but no official driver). The ESP32 also has a lot more memory on the chip which can be important if you're using it to retrieve data over the internet; it's not that important if you're sending packets but if you're using HTTPS, that uses quite a bit of RAM. It all comes down to what it's used for.

In my uses for the Pi, an ESP32 is usually not viable, as I want a full OS with a lot more resources available to me. It is, however, interchangeable with the numerous other SBCs that run Linux made by other companies. I haven't needed to buy one recently, but it is quite possible that the next SBC I acquire will be from someone else.

Server broke because it was invisibly designed to break

I think some of your critiques are making assumptions that aren't necessarily the case.

"If this is a mission critical application, why didn't "Kris" have a fail over contingency plan?"

We don't know how critical it was, just important enough that it being down for days led to some unhappy people. Since they weren't threatening to fire him, just putting pressure on, that suggests it was probably important to someone but not mission critical.

"No mention of a fail over server or cloud based solution to keep the application going."

We don't have an exact date on this, but it appears to predate the availability of cloud, so it would have to be another server or a different one with extra capacity. Another server isn't cheap. It's not automatically his fault if the request for that level of overprovisioning was declined.

"Why wasn't there a service and support plan on the server?"

You assume there wasn't. Someone got called out, didn't they? Why couldn't that have been as a result of a support plan? They could have one which guarantees a support person shows up (which happened) but doesn't automatically cover all expenses, hence why they would expect to pay afterward, especially as there were new parts brought in as a result of the failed attempts.

"Since they pay for each repair call, why doesn't "Kris" have some spare parts on hand."

You assume they pay for each call. Perhaps they only pay for parts, and their support plan requires that they only use authorized spares. Why buy a ton of those if the engineer is supposed to bring them? Also, how do you know they didn't have some of those things. Maybe they replaced a broken power supply with their spare a while back, hadn't received the replacement spare, and then the engineer decided power supplies had to be replaced (for no good reason as you've already stated) and made them get two new ones. You can assume any level of competence you like, and assuming the lack of it doesn't prove it any more than assuming they planned for everything.

"Where I have issue with "Kris" is where he blames the repair tech for a wrong diagnosis and the amount of down time."

Both being the tech's job. The tech should try to diagnose things rather than just swapping out parts every three days until it worked. Had the tech tested each component, leading to a day's downtime while he went through everything he could think of, I don't think there would be that much complaining as the downtime was necessary to identify the problem. However, the tech's approach didn't appear to check the hardware very much, so most of the downtime was waiting for stuff that wouldn't be needed if the tech tried diagnosing the cause instead of guessing.

Corporate execs: Get back, get back, to the office where you once belonged

Re: Oh good, more dumb absolute takes

I agree a lot with this. The arguments over WFH appear to break into two categories: the people who like working in the office (or the managers who like to have their managees there) insisting that remote work just doesn't work, and the people who like WFH and insist that there are never any downsides to it or advantages of an office. I've seen nobody even try to analyze the possible benefits to the one they don't like as much, just a bunch of insinuations of how stupid and lazy the people who take the other side must be. Then again, it's definitely not the first time when it would make sense to analyze how things really worked and people found a reason not to bother with anything that could be objective.

I've worked in an open plan office. If everyone came in, the noise and chaos were disruptive and I wanted to work from home. If nobody came in, there was no point being there so I wanted to work from home. If a few people came in and they were people I worked with, things worked, although there was still some potential of disruption. If a few people came in but weren't the people I worked with, it was pointless again.

I've also worked where there are offices, and if everyone came in then it helped with working together and didn't create a disruptive environment. Walls kept the noise from a million conversations from drowning out my thoughts and permitted me to have a meeting without necessarily disrupting everyone who worked in the area. More people chose to come in when there were separate offices. Imagine that.

If you're a manager who wants people in the offices, consider whether people actually work well in the offices. In both cases, there will be some people who just don't want to come in, but there will be a lot more take-up of the idea if the offices aren't unpleasant by their design.

Qualcomm talks up RISC-V, roasts 'legacy architecture' amid war with Arm

Re: No better time to start a new mobile OS

Then try changing the compiler target for those existing Linux versions to RISC-V and build a device around it. It won't do any better than ARM versions did, because the average user, and even the technical user when not building the device, doesn't care what instruction set is in there. A phone with RISC-V and a phone with ARM look the same to a buyer, at least if the processors are similar in performance and power consumption (which is probably not the case right now anyway).

Just because a part is undergoing a major change doesn't make it any easier to replace other components. The efforts to build a better mobile OS, which I wholeheartedly support, will not be boosted by a different ISA. We'll have to build them the hard way, which means that we'll get the same results if we stick with ARM and cross-compile later.

Apple preps for 'third-party iOS app stores' in Europe

"weakening the ecosystem safeguards to allow non-curated apps onto iPhones provides many new and interesting exploit vectors for miscreants to discover."

No, it doesn't. You don't need Apple's permission to load code you right onto your phone. That's all you need to search for and discover exploits. An alternate method of installation doesn't let you run anything that you couldn't before; it lets others more easily install something you wrote. This won't generate new exploits, but it would allow someone to try releasing their exploit-laden app outside Apple's review systems. Of course, if it's a working exploit, that app would get through the review anyway, and Apple would still have to fix their OS, not their app installation method, to deal with it. The security systems in place that sandbox apps would not have to be weakened and exploits would not be any easier to find with this method in place.

Re: I hope there's an option to block this nonsense.

Someone doesn't know how to use a computer, but can make an appointment online, has a smartphone, and knows how to enable a different app store, even through Apple's security warnings, but can't read instructions? Those factors are mostly incompatible with each other. By the way, if they know how to mess with their app installation methods, they can also figure out how to use the browser that comes installed on every smartphone, so their lack of a computer is not a factor. This person would be confused how to do anything if they're unaware of computers and can't read the instructions they were sent. In this case, there is a big problem, but it's not where the app is listed. It's that someone incapable of following a process they need was sent only one option for doing so, an option that relied on them having hardware they might not have, which would be a problem. Fortunately, that's not how such things have been done, and this is a made up example you're using to make a flawed point.

Re: Apple could...

If your friend uses local backups, they might have an older version coincidentally backed up which they can add. I'm not sure if that is portable across devices (I'm assuming not), and few people bother with iTunes syncing or backups, but it could be worth a try.

Re: I hope there's an option to block this nonsense.

"Which app store do you think those daily-living apps will launch on within the EU? Apple Store? Or only on the “European competitor” App Store? You know the answer."

I do, but it's not your answer. Governments have been happy to put their apps in the normal stores forever. Let's look at Android. Which store has the government apps? Is it FDroid? No. Is it EUStore? No, that's not a thing. It is, in fact, the Play Store where they put them all. So I wouldn't expect them to want to build another.

Let's say they do anyway. Given the level of success such things have had before, I doubt that would ever happen. Let's assume, though, that I'm wrong. So what? If you find you need an app that's not in the normal store, you can either live without it or accept enabling that store just to download the app you need, then not using it for anything else. If it works anything like Android, you can enable the store for the app you decided you want, install that app, then disable the store and the app still works.

Re: I hope there's an option to block this nonsense.

I've got a magical solution for anyone who doesn't want to use third-party app stores. It's a bit complicated, but I think anyone reading here can manage it. Just follow these steps:

1. Don't turn it on.

Seriously, that's all that's involved. They're not mandating that every possible store is preinstalled. They're not providing for automatic installations. What they've asked for is a switch which Apple can bury in the settings, put scary security warnings around, and which you have the freedom to leave turned off. Nobody would be required to sideload or to use anything that Apple doesn't run.

Re: Thanks for nothing EU

Then don't list there. Problem solved. Everyone will have the Apple version preinstalled. Everyone who uses apps will have some that you can only get from there, so everyone who wants your app will check for it there. This is the same as on Android. I get a lot of apps from FDroid, but if I need something specific, I already know that you can only get it from the Play Store, so I get it from there (via an anonymized connection, but it still comes from there). People who don't want to release as open source don't have to use FDroid for me to know how to install their app.

Here's something communism is good at: Making smartphones less annoying

Re: "but there is still a difference between a state that intends to organize itself to loo"

The difference isn't in how bad things can be. Some of the worst dictatorships have been communist, and some have not. The difference is only in the appearance and structure, with some related effects on exactly how badly it is run. One major appearance difference is in the structure of the economy, as communist countries tend to talk a lot more about the ostensible strength of the worker than non-communist dictatorships, organizations tend to be legally depowered instead of just practically depowered by government action, and many organizations would be state-owned instead of private. This doesn't make them better or worse. It's like the difference between desktop environments on Linux; they look different and have different effects on the user's workflow, but they all basically do the same thing.

China under Mao spouted a lot of the classic communism points and set up some of its trademark social programs [complete fiascos]. The Soviet Union codified a lot of those things originally. Cuba and North Korea still sound like that today. Non-communist dictatorships often sound very different, if no less menacing. China today, though, is not structured like it once was and no longer uses the classic indicators of communism. It has adopted a less communist appearance, but is no less authoritarian than it was. They switched off "communist theme" on the UI, that's all.

Re: Is it even Communism

There are differences between communism-themed dictatorships and other types of dictatorships, yes including fascism-themed ones. We've never seen a non-dictatorship communist state, and I think the evidence is clear that such a thing isn't possible, but there is still a difference between a state that intends to organize itself to look (or even to be) communist under the top echelons of the dictator and his supporters and those that don't. China used to look like the former, and it was really bad. They don't look much like that anymore, and it's still bad.

To protect its cloud, Microsoft bans crypto mining from its online services

Re: Over selling

"Not that I think it would be a sensible thing to do, of course - but bitcoin is just a compute load. How is it any different than someone who wants to compute Pi : it's the customer's decision as to whether the service is worth the cost. If the provider is making that decision then they've underpriced on the assumption you overprovisioned and they want to hide it."

Treating mining as any other workload seems to have been the policy for a while. That they've accepted people mining on provisioned resources for years suggests that it was probably fine. This means one of two things:

1. They didn't underprovision for years, but they decided to start doing it now. Doesn't seem likely they'd change now.

2. They have another reason. The article suggests one: they're seeing miners stop paying the bills because the value proposition has changed for them. As those miners would have rented some of the most expensive boxes available, those would be some large unpaid bills. I can see why taking some losses and looking at having to file a lot of claims to recover unpaid bills would convince someone in finance that maybe we should cut off this kind of work entirely.

"coin mining is hopelessly uneconomic on a general purpose CPU"

Correct. Basically anyone who is intentionally doing it is using the VMs they've got with lots of GPUs attached. Those cost more, which is probably why Azure has accepted miners before, but with the crash in prices, they've probably had some people not paying their bills as suggested in the article.

Meanwhile, if someone is doing CPU mining, then it's probably that someone is using unauthorized resources, from an external attacker having found a way in to an untrustworthy admin who thinks the increased CPU usage won't be noticed.

The IT decision-maker that really matters? Your pet

Re: Why?

That's not it. The video vantage point from wherever a pet chooses to be isn't a great one, and our lives aren't interesting enough to justify the processing required to get anything from that. The amount of useful advertising information to be gained from the effort is too minimal.

The people making this product available aren't supervillains planning to sneakily bug your house. They're people saying things like "Hey sales, do you think you could convince someone rich who has a cat that they need extra sensors to indicate that their cat is sleeping on a soft surface in a sunbeam? How much do you think we can charge, and how quickly could we break it and have the customer want to come buy another one?". It works.

Voice assistants failed because they serve their makers more than they help users

Re: Business model is easily adaptable for Amazon

You missed a zero in that: it's $50 per speaker per year. Moreover, if they introduced a subscription, that's more people who will dump the thing in a drawer and not use it again, as now it's a brick unless they pay again. I'm not sure where the money went, but maybe they should start looking at the logs they undoubtedly have to answer the question "What is the stuff we spent a ton of money writing scripts for that nobody asks for".

Re: Not trusty

"If a machine pretending to be a human is to be successful, it should at least behave like the sort of human that people can trust"

I don't think this is their problem. Something that sounds like a human, could do things, but also slipped in some advertisements would probably do much better. Right now, the problem is that the machine isn't pretending to be a human. It refers to itself in the first person from time to time, but it acts robotic, doesn't remember anything, and gives the same scripted responses to almost everything you could say. Talk to one for three minutes and it becomes really obvious how not human it is, and talk for another five and you learn that it's not a robot that can actually do anything very useful. I'd like it if people automatically rejected such things on privacy grounds or even had a subconscious feeling that something's manipulative here, but in reality, I think they just discover that the device is a curiosity at best and that they don't need one after all.

I think you have a better idea of this than the article does. I don't have any devices intended for voice use, but I do occasionally use the voice interface on my phone. Here are the commands I issue:

1. Call [person].

2. Set a timer.

3. What will the weather be today.

4. (rarely) Send a message to [person].

The first one is designed to save time finding the person in my contact list. The second and third are similarly reducing the time and mostly used when I'm doing something else with my hands. And that's all I need it for, meaning a tiny model that recognizes pre-defined phrases would be just as capable as a conversational bot with thousands of scripts written.

There is one aspect where the obviousness of the manufacturer's influence might cause a problem. I know someone who has an Alexa device and occasionally asks it to play music, but they don't have whatever Amazon's best music service is. This means that, when they ask it to play something, the device usually responds by informing them that they can't get it, but they could if the user buys a subscription. It will then read out the standard information about the subscription, such as that there is a free trial, how long it lasts, that it would renew automatically, and what the cost of that is. You have to wait for its thirty-second speech to end before you can tell it no. I'm not surprised that this person has put the device in a drawer and never turns it on now.

Re: Alexa:

The phrase "pod bay doors" does not appear to be in my lexicon.

Fine, I had to modify the quote a bit, but it's a lot more likely to be what Alexa would say. It was originally "cargo bay doors". I'm going now.

Re: "... they serve their makers more than they help users"

"while having a conversation with my brother by phone, and without saying the key word to wake the darn device, 'Alexa' just chipped in with information about the topic of conversation without being prompted. So anyone who thinks these don't listen and process except to pick up the key word need a wakeup call: They're listening, processing, and can/will respond without the keyword ever being uttered."

I don't like or use these devices, but you've jumped to the conclusion that they're sneakily listening in when what actually happened is that something you said sounded like the wake word to the basic listening software. Sneaky listening devices don't start answering questions you didn't mean to ask, but devices that incorrectly think that you tried to activate them do. There are privacy concerns, and theoretically their manufacturers could turn them into bugs when they want (research has demonstrated that they're not doing so right now, but that's no guarantee of anything), but your experience has caused you to mistake a bug for evidence that it isn't.

"the one feature missed is the mic 'off' switch."

It's a button on the top of each device. Press that and the microphones are disabled. If you are afraid that Amazon's designed them to lie about it, then maybe you have a point (people have checked it in teardowns but I'm not one of them so won't promise anything), but if you distrust Amazon that much, maybe you shouldn't have them at all. I don't trust or have them, and yet I expect that the button to mute the microphones will do exactly that.

Raspberry Pi supply chain loosens just in time for the holiday season

Re: Would have preferred Pi 5 announcement

Yes, I am. The kind of semantics that are critical in tax and employment law. Whether someone is a contractor is a tricky issue that has caused problems in many countries. There are incentives for the company making the profit from the work, the worker and any instruments they've created, and the tax authorities to classify people as a contractor or an employee, and such incentives are often very confusing. It's how you can get two people doing the same thing under basically identical circumstances, one of whom insists they must be regarded as an employee while the other insists on being regarded as a contractor, and that's only the first of many disagreements. Complicated issue, that. Selling a physical item that's not even restricted (if I buy a bunch of BCM2711s from Broadcom, they'll be the same as what the Raspberry Pi people have) is a lot less complicated from all perspectives, and hence a comparison between them fails to make any point.

Re: Would have preferred Pi 5 announcement

I'm neither an expert on tax law nor work in the UK, but I'll point out that there is a legal difference between "provide service" (grey area), "work for" (what IR35 is ostensibly meant to deal with), and "sell goods to" (the situation between Broadcom and Raspberry Pi). I've heard enough to think that IR35 is wrong a lot, but the comparison you're making is still invalid.

Re: Would have preferred Pi 5 announcement

"Not long ago they had the best SOC for the money."

I'm not convinced about that. There have been more powerful chips used on SBCs for as long as it was obvious that people wanted to buy Raspberry Pis. I'm not sure how you decide what performance and price intersection is right, but their competition offered faster chips at all points. Until the Pi 4, their chips were pretty uncompetitive. Even when that was new, depending on workload, they were still uncompetitive (for example, RasPi's SoCs have always omitted encryption acceleration, but nearly everyone else's products support it). They have been more consistent about being inexpensive and easy to buy (supply chains excepted, but lots of local distributors and not having to pay half the price in shipping that takes a month).

Raspberry Pi has never really tried to be the fastest board out there, and it's served them well. I wouldn't expect them to change that just because some new boards show much better benchmark numbers. They can correctly point out that those boards tend to cost more and use more power than theirs, that the software landscape for them is more chaotic than the Pi's, and that they don't need to change their design in order to consistently sell out of the models they do make. I don't think they're aiming for the perfect board, so their next version is unlikely to meet all your or my expectations. Fortunately, we do have multiple other options to consider. My last SBC purchase three years ago wasn't a Raspberry Pi, but it still works.

Re: Would have preferred Pi 5 announcement

Raspberry Pi isn't going to adopt RK3588 just because it's faster and more efficient. Among other things, they want images to be compatible with every board, which is easier when using basically the same architecture and is not compatible with a SoC whose boot process is so different. Also, they've probably got a good relationship with Broadcom given that every board has had a Broadcom CPU and GPU on it. Abandoning that in favor of a competitor is not likely. It's also worth keeping in mind that the Pi has never aimed for the fastest chip out there and has frequently lagged other companies on raw performance. I don't know when the next Pi comes out although I wouldn't expect it to be soon given the production difficulties, but it's not likely to give you every feature you want.

If you want the RK3588, there are several companies who build SBCs around it, some explicitly designed to be similar in size and peripherals to the Pi. You can have those now and many have more availability. Don't hold your breath for the Pi 5 being announced when you want it or having every feature you think it should.

OK, we know iPhones are expensive but... $11 a month for Twitter Blue on iOS?

Re: Apple users can afford it

"Rhe original example was £200 for 8GB of Apple storage. Some rip-offs are bigger than others."

Read again. That was for 8 GB of RAM, whereas this was for 256 GB of storage. RAM costs more per GB than storage does, whether marked up or not. Not comparable. Their mobile devices don't offer different RAM options in any case, and I'll admit I've seen some laptop companies with similar crazy markups on RAM and some that appear to use numbers closer to real market prices for the stuff. Apple does have crazy markup for both upgrades, but they're far from the only ones.

San Francisco investigates Hotel Twitter, Musk might pack up and leave

Re: Not unusual

"We don't know what their schedule consists of or what their arrangements are (at least I don't)"

I don't know the specifics, but I don't think it's that hard to connect Musk's vague statement of "long hours at high intensity" and the fact that they brought in beds to arrive at a rather grim picture.

"They might be putting in more hours and getting paid more"

They might, but I'm pretty sure they're not getting paid more, given how eager management have been to fire everyone else. These workers are mostly earning salaries, which means no overtime. I haven't seen any reports of a round of raises or bonuses. Have you?

"they might be working more productively,": I wouldn't count on that. Denying people work-life balance tends not to end that way.

"or they might be accumulating their time off and taking longer weekends,": Again, salaried workers don't get time off in compensation for long hours. I'm sure they are saving up what they normally get because they have no other choice, at least until the company switches to a model where they can no longer accumulate any (just a prediction, as I haven't heard them plan this yet).

"And if they are putting in more hours, they must have some incentive to do so, otherwise why would you do it?"

I don't expect many to keep doing it for long. As for why you would do it, you would do it if the job is necessary for something you rely on until such a time as you get another, hence why many of the employees are on visas that make it difficult to simply switch jobs.

"I'm sure you wouldn't mind offering appropriate references to back up these assertions:"

It's not like every plan of Musk's is fully documented in public. Most of it has been leaked in vague terms. I can't give you a direct citation for each thing, and you know that well.

* That's not why the beds are there

Derived from the demand for long hours, which is compatible with having someone working all day and sleeping at the office to make sure they're working when they would have been commuting, but less compatible with letting employees take a nap during a normal work day.

* They're there because Musk thinks coding volume is his problem

Musk wants developers to work much harder than they are because he thinks that will solve problems. This despite the fact that Twitter's problems are not due to missing features or the like. It's pretty clear he doesn't have a great understanding of what caused Twitter not to make a large profit, as focusing on things that generate revenue would do more than insisting on more code written.

* doesn't care about the long-term viability of the people he wants to build the systems

Long hours at high intensity does bad things to people. It's called burn out. Anyone who actually goes along with the demands for a long time is likely to find this out. He has the intelligence necessary to understand this relatively basic concept, but he doesn't care.