Posts by doublelayer 10681 publicly visible posts • joined 22 Feb 2018
Is this supposed to be a joke? I'm really not getting it.
Imagine burglar goes on the usual sweep. Now they can take a selfie at someone's house, then use AI feature to remove themselves from the selfie.
Then when Police catches them, they can show them the picture "Look guv, I wasn't there!"
Look, you have a picture of the crime scene at the time of the crime, on my phone. I feel like there's something in your proposal that I'm missing. I'm going to need more explanation.
It's about what you can do after learning to use one tool. If everything is different, you know how to do the one task that tool is for. If things are standard, then you can probably figure out how to use other ones. So, for example, if you need to use a new piece of software for a bit, you already know the basics of how to open it, how to close it, where the settings are, what stuff you can click on and what stuff you can't, and you skip right to figuring out what you need to do inside it. If you might have to relearn the concepts, it's a lot more frustrating. For example, if you're used to ctrl+c copying and ctrl+v pasting and you also have options in the edit menu for those functions, but your new program has no keyboard shortcuts and the buttons are revealed by clicking a toolbar button with an unfamiliar icon, you spend more time learning that.
People who frequently have to pick up new applications will find this annoying. I imagine you've experienced it at least a couple times yourself. The worse situation is for those who don't do this frequently. How many people have you met who resist any change. An update might move one button and they panic? That's often because they've experienced something which completely changes their workflow and they need to spend a while re-learning things for no reason. There are times when that is necessary, but generally, if people already know some paradigm and yours doesn't offer a specific benefit that you can concisely explain and prove correct, it's best to stick with the working one. This is a lesson that many companies, especially Microsoft, would do well to follow. I wonder whether people would stick less to old Windows versions, even as they go out of support, if the next one looked familiar.
That claim requires some evidence. I'm sure you can use vi quite efficiently, but is it automatically the case that it's more efficient than an experienced user in a different editor? I think that depends a lot on what the user is trying to edit with it.
I know the basics of using vi for systems that have nothing else, but I came along when text editors could do more things and so I've never taken the time to use vi instinctively rather than deliberately. It's perhaps unsurprising that I find it to be slower to use than editors I've spent lots of time in. However, I also think that may be true if people find certain features to be useful. A typical GUI editor tends to have more builtin features than vi does, partially because there are other programs that can be used to get the same effect on the CLI. The more of those the user finds useful, the faster the editor with them included will be to them.
It is certainly an issue, although your example doesn't work very well:
"Imagine being a teen interested in electronics, but not even having space to store cereal away properly."
Electronics isn't too space-intensive as hobbies go. The person who doesn't have extra space will have to pack up their tools, which is a bit more annoying than keeping them ready for use, but they'll still be able to take them out, do some experimentation, and put them away. There are several hobbies that are much harder than electronics to fit into a couple drawers and a box which might have suited your argument better.
It is at least partially the other way around: Mozilla wanted to have better tools, so they made one. Others then took it over. It's embodied the successful aspect of open source, because people improving Rust also improve a tool that Mozilla relies on. Unfortunately, it doesn't make the negative aspects of a lot of people burning out go away.
Maybe because you're both assuming that the speedup was both completely unnecessary and wasting time. Neither is proven, and if we believe their statements, both are false. Even if we don't believe them, both could easily be false.
Why write a replacement for an API at all if it's already fine? It sounds like some speedup was necessary. If you're writing a replacement from scratch, then doing it efficiently the first time is the best policy as long as it doesn't result in taking a really long time. If you write it to be faster, just enough to meet the number they provided, then that number will eventually increase and you'll have to write it again. If you significantly exceed that number, you may never have to do that. There is a possibility that this did result in wasting time, but if the manager said "we don't need the speed" rather than "you didn't need to spend so much time", it doesn't sound like that was their complaint.
And that you could do so was obvious from the start. If I or anyone else with an open source project want to make it impossible to make money from it, we can manage it. We don't want to prevent that. My measure of success in my open source code is not "I'm sure nobody can make a profit using this". If someone manages it, that's fine with me.
Part of it is defining all open source work as an unpaid internship. It's not, and we all know it's not. I've been employed full time for quite a while now. I'm no longer doing internships, but I still contribute to open source. If every line of open source was being written by people who had no other experience, the quality of it would be a lot worse.
The other problem is trying to pretend that open source work is a requirement to get a job. Yes, it can help to prove experience, and people will take that into account, but other types of experience work as well. I was not required to show open source contributions to get my first jobs in programming. Since I had less experience, it took more work to prove that I knew what I was talking about and I got more rejections, but I and many like me still managed to do it.
Two incorrect statements about what open source is and how it works don't help make a point, especially when the point is unclear and slightly insulting. As far as I can tell, the problem the rant is trying to point out is that open source software is written primarily by people who have the time to do it well, and those people generally already have money so they can afford to spend the time, and I'm not sure what the solution to that was supposed to be. Yet, by not doing whatever we were supposed to do, working on open source is propping up some system of racism and sexism by corporate proxy. It's a garbled complaint that relies on incorrect assertions.
And Java was not open source and still isn't. There are some open source versions like OpenJDK, but the Oracle and Sun versions are not. There are lots of open source language tools. GCC, for example. How much to buy that? Or databases, how much would it cost me to buy PostgreSQL? There are companies that make a product that you can see the source of, and they can be for-profit companies, but most projects are not run like that and, even if they are, my choice to contribute to those projects is not the same as getting hired by those companies with a salary of $0.
"you can't actually decide to work for free at for profit organisation (with some exceptions). It's illegal."
Which you will not stop saying until you eventually recognize that open source work is not for profit. Someone might take it and make some profit of their own, but my work is not for them. It is for me, or the group of developers (not an organization at all, let alone a for profit one), or for the organization that organizes to write it (probably a recognized nonprofit organization). Someone else making money does not mean I am working for them. If they tell me to do something and I don't want to do it, I don't do it, which is a pretty different situation from a working environment.
The types of people who work in open source are the result of a lot of different factors, some of which are important to do something about, but none of which will be fixed if you keep telling us how the entire concept is or should be illegal for incorrect reasons. Your reason, which is not the only one, is that some people have more money, and therefore more free time, to spend on their choice of activity. Of the various reasons why imbalances exist, this is one of the least important. Differences between who gets related education and cultural problems in some projects resulting in people feeling unwelcome are both more important problems and, crucially, problems we can actually try to fix without causing more damage than we avoid.
You have demonstrated that you don't know what open source is (it's not for profit, for example), nor why people do it. People have the right to decide to do things for free. Some people choose to. Others may not have the freedom to spend their time doing it, but the same is true of literally anything that takes time; if you already have what you need, it is easier to have the time to spend on hobbies whether those hobbies are writing open source software, baking, painting, or walking outdoors. What you're saying effectively boils down to "you have free time and someone else doesn't, and that's a problem", but open source is not special in that regard.
The legal agreements on that software are not at all ambiguous about you needing a legally purchased Apple computer to be allowed to run it. You may be able to break those agreements with sufficient effort, but it is not what you're supposed to do any more than I'm supposed to break the Windows license check system. So no, it is not free as in beer or in speech.
Try again. First, you know Apple reduced to 15% relatively recently to try to counter accusations of monopoly behavior, right? And you know that Google followed suit very quickly. From 2021:
Google is lowering commissions on all subscription-based businesses on the Google Play Store, the company announced today. Previously, the company had followed Apple’s move by reducing commissions from 30% to 15% on the first $1 million of developer earnings. Now, it will lower the fees specifically for app makers who generate revenue through recurring subscriptions. Instead of charging them 30% in the first year, which lowers to 15% in year two and beyond, Google says developers will only be charged 15% from day one.
The company says 99% of developers will qualify for a service fee of 15% or less, as Google is also further reducing fees for specific vertical apps in the Play Media Experience Program. These will be adjusted to as low as 10%, it says.
However, emulation of X86 Windows is a bit easier on ARM Windows that has all the same calls as X86 Windows, rather than running through Wine which has some gaps. If the emulation continues to exist, a lot of things can be used if X86 dies. Of course, given the performance differences between current X86 and ARM chips, I wouldn't expect X86 to be dying any time soon.
The same reason that people try to run Linux on them: they want to do some Mac OS stuff and some stuff in a different OS, and they don't see why they should need two computers to accomplish it. It works for my laptop which runs Linux, multiple versions thereof, and Windows and I can pick whichever I want at the time.
Do you have a source for the Russian military not having any communications equipment? It doesn't seem that plausible since they have a lot of equipment they've been burning through. Radios are not that expensive to make and don't get used up as quickly as weapons. Meanwhile, they keep blowing up infrastructure including electricity and communications systems, so if they were using their own phones, they'd frequently drop out. They can also buy radios from any number of Chinese manufacturers who make radio equipment all the time.
And, even if it were true, Russia could also get a lot of cheap Android phones for their military if they needed them. I'm not seeing facts here, but supposition and wishful thinking. It would be great if Apple could push a button and throw the Russian invasion into chaos, but they can't.
I'm not sure that bricking iPhones would have much effect on the war. Paying the fine will, but most iPhones are in the hands of consumers, not government leaders who have their own phone system, at least purportedly, and could easily switch them out at little cost. There's a good reason for them to refuse to pay the fine. There's not a great one for breaking phones that have already been sold.
As someone who didn't downvote you, or upvote you, the title suggestion is a distinct possibility, as it's unrelated to the issue and people disagree with that sentiment too, but it's probably not the only reason. Another one is that there are some places that are truly disconnected from all grids, generate their own power, have no old copper phone cabling, etc. It is not most of these places. Most of them do have some infrastructure connected to a house, but good internet is not one of them.
There are some locations that are so disconnected from everything around them that it really doesn't make sense to provide a cable to them, but most of the people they're talking about live in small rural communities that don't have good service, where there are already cables for power and where a local hub could connect many people relatively cheaply if the work of installing that hub is done. The existence of completely off-grid houses doesn't make them the majority. Nor are those the priority for this program, as the program is intended to lower prices for areas where the poorest live. Chances are that if you're using a bunch of solar panels to power your off-grid house, you've probably got more ability to pay for your own Starlink connection than most of the people who don't have broadband.
It's not about return on investment. If they wanted to do that, people in rural areas could continue to live under the "we don't earn enough from your area so have fun with the microwave link" policy. They've gotten money to add service even though it won't turn a profit, and they're trying to use that money to add service that will last for some time. They could pay Starlink, and a lot of people would get service right now and the service would, while not meeting the speed requirements most of the time, be quite a lot better than what the users had before and useful for most things. The problem is that the money is not a recurring annual budget, but a one-time allocation. When it runs out, the satellites are still as expensive as they were before, they still need replacing every five years, and so there will be a lot of ongoing costs which will return to the individual consumer. With cables already having been installed, the fixed cost will have been paid for and the ongoing costs will be lower, so there's actually some chance that people can continue to run them after the money has been spent.
They don't have very much experience, but what they do have is a lot of people, a willingness to see them die, and a lot of organized training. Most of the population is either in a long period of military service or has already gone through it, and while that service mostly included being used as really cheap construction and agricultural labor, it also included training on how to attack things. I'm sure they'd lose many of those people if they started to use them to attack, but they won't be too worried just because thousands are dying unnecessarily. They also have some more powerful weapons to resort to when they inevitably start failing. Those weapons may not be plentiful, but they are serious. So it's really best for everyone if we can somehow prevent them from starting a war at all, even if we would eventually win.
They probably aren't using one directly, but a local VOIP provider. It might be a giveaway if it's all coming from international numbers, but if it's a number from that country, there's less that can be done to identify whether it's being proxied for an international origin. In addition, it's probably not being proxied in this or many other cases. The people who want to attack a US election official are probably in the US to begin with, so wherever they're calling from, it's in the country. You would have to use geolocation data to identify where the caller was, and I don't think the systems for collecting and reporting it are fast enough to let them filter it. With all of these obstacles, they probably send the calls to the local authorities rather than deal with the risk that someone actually has an emergency and they ignored it.
"Whatever happened to being put on trial before a jury of ones peers."
If he ever ends up in Australia or a country from which he might be extradited, that's exactly what they intend to do with him. It's always been quite normal to name people you intend to charge but can't arrest because they're not available, ask for information related to their capture, and put restrictions on people dealing with them. If he turns out to not be involved, those moves will be rescinded, but your contention that this is somehow new is incorrect.
You may have misinterpreted my comment. I did not say whether I think cryptocurrency is a good currency. For many of the reasons you stated, I think it is a bad one. However, I wasn't rating the four options (cash, banks, CBDCs, other cryptocurrencies) as best currencies, but only on the "how easy is it to take from you by force if you assume that the force concerned has complete legal authority" index. I don't think that's a very important index, but it was the point others brought up, so I was using that metric.
I don't have or want any cryptocurrency, but some of your criticism isn't global to the concept. For example, the energy requirements and privacy problems are both entirely accurate descriptions of Bitcoin and many ones like it, and both are big problems to making them useful for anything. However, they do not cover all currencies. There are privacy-focused ones like Monero which don't make it easy to track transactions the way that Bitcoin does. Monero is also easier to mine on commodity hardware, which is usually considered a bad thing because it's what cryptomining malware tends to be mining. I'm not sure how well Monero scales because, since I have no interest in owning any, I haven't bothered to learn all the details. It's possible that it would also fail badly if it became more popular. However, it's useful to keep in mind whether the complaints you have are intrinsic to the concept of cryptocurrency or not.
For example, the dodgy exchanges are a big problem, but that's a legal thing. If some government chose to regulate them like banks, including mandatory insurance programs like the FDIC does, they could have similar security to banks. The exchanges don't want it because they want to get as much money as fast as possible, but they could be required to take those steps. Whether the companies handling the commodity are trustworthy isn't related to what they're storing, but the oversight of their operations, regulations they must follow, and their accountability for failure to follow them.
In their defense, there are significant differences between the options which make one harder to take than the other. If you are concerned that someone will use legal authority to take your money, some systems would be preferable to others. I am not concerned about my government doing that, so it's more academic to me, but the differences aren't minimal. My reasons for opposing CBDCs are more related to privacy and security than to fears of confiscation.
For example, if we assume the willingness to use violence to get your money, then any system can allow it to happen, but systems that require individual attention on you would add some difficulty, and systems that make it feasible to hide some value make it more likely that you can avoid giving all of your money to the attacker. Cash and non-centralized cryptocurrency have those abilities. A CBDC probably doesn't, depending how it's implemented. Modern banking systems do not have it very much.
The conversation included discussions of physical cash and of banking systems. I agreed with you that physical cash is relatively difficult to confiscate and stated that money in banks is no harder to confiscate than a CBDC would be. I think my use of the terms made it clear which one each statement was about.
For the average person, the distinction is important. Most of my money is in a bank. Most of my payments either access the bank directly or through a credit account. Thus, my money is available to be confiscated by the government. I don't store much money in cash form, nor can my important bills be paid, either easily or at all, in that form. Most of my bills don't have a facility to take payment in the form of cash, nor does my employer offer me the ability to be paid in it. Thus, should my government decide to attack my assets, I would be vulnerable to that attempt, CBDC or not. Of course, I could withdraw my money for physical cash to get around some of that, but it would not work forever due to the previously mentioned reasons.
You also appear to be misinterpreting what a straw man is. I did not argue that physical cash is vulnerable because banks are, in fact I specifically stated otherwise. The discussion as a whole, however, is about CBDCs which replace banking systems and possibly, but not certainly, physical cash. Thus, discussing the privacy differences between private banks and a CBDC is relevant.
That's true of physical cash, but not so much of existing banks. If the government wants to freeze your account, the banks can do it for them. There are some legal impediments to doing it, as I'm sure there would be for a CBDC-based system, but I wouldn't expect them to be that much different between the two systems. A CBDC system that retained physical cash would probably be no easier to dismantle than the current one with private banks.
That survey has a weird definition of "elite", as well as a lot of other problems. For example, one of the strict criteria for both is that they must live in dense urban areas. If you're a rich person who has enough money to afford a large house, you'll be excluded from the elites set by definition. The entire thing is designed to give a predetermined answer. This can be demonstrated clearly with one of the summaries of the survey questions:
About six of ten elites have a favorable opinion of the so-called talking professions—lawyers,
lobbyists, politicians, and journalists.
To anyone who thinks about it, it's probably pretty clear that people have a different opinion of journalists than lobbyists, and probably have a different level of regard for politicians they agree with and ones with whom they disagree. The survey, however, lumps these all together in order to misconstrue their answers. I am pretty sure the same problem will be found in many of the other questions as well. The sample size, by the way, is one thousand people chosen online between the two sets, which is ridiculously small for any survey. This survey is not a reliable source of any information about who thinks what.
Mobile oses aren't synonymous with kernels. Tizen is not a kernel; it's using Linux. WebOS uses Linux. The only one on that list that doesn't have Linux at the base is QNX, and that's the one that's no longer used on basically anything.
Of course, anyone can write a kernel if they put some time into it, but it's much harder to write a kernel that has anything like the ubiquity and stability of Linux. That takes a lot more work, and it's unlikely for any particular company to be motivated to do that work when using stuff that already accomplishes some of the goal is free. They can prove us wrong, but I'll believe that they have when I can prove it for myself.
You're rounding a lot to turn 1.4 billion into 2 billion. Yes, it's a lot of people, but you can't just assume that, because there are a lot of people, you will get a lot of them to buy your products. Do I think that Spotify would make a new app for Chinese users? No, and here's why. From Wikipedia:
"It [Spotify] has no presence in mainland China where the market is dominated by QQ Music."
If they haven't entered the market now, when their Android app would run on most devices as it is, why would I expect them to do so when they first have to make something completely new? They have decided so far that the likelihood of success is low enough to focus on other places. The same is true of a lot of companies outside of China.
The Harmony OS for IoT devices is not at all the same as the one for phones. Maybe the new version will find a way to unite them, but until now, they've been completely separate things with the same name on them. There was no compatibility between the Harmony OS watch and the Harmony OS phone; you could not install an application written for the phone on the watch and not just because the storage was insufficient. This is not really a surprise, as the Harmony OS watch is an embedded system running one program close to bare metal with the embedded Harmony OS providing features. Lots of things like that exist, and they don't tend to run Linux either because they run in environments with insufficient resources to make running Linux possible or desirable.
However, we don't take any of the embedded OSes for low-resource electronics and try to run them as the main OS on a phone or computer because they're missing lots of features and there's no point. The amount of code that can be usefully shared between the two is too low, because the low-resource version will need lots of optimization to fit into the constraints, whereas those optimizations generally just make things worse for the kernel that can use more RAM and should to keep the user experience fast. Can they be united? Definitely, but there's a reason to believe that whatever Huawei is doing, it's not that.
Hence the "breaks maintainability" part. But phone manufacturers don't tend to get too much PR problems when they never update Android, hold back security updates for a year, for the first two years, then stop entirely, or run out of date kernels. Chinese companies are among the worst for keeping them updated, even nowadays when companies like Samsung have realized that Android support longevity is important to at least some people. Huawei is one of the companies that still has no organized update policy. Do you really think the public perception of them would be much worse if they renamed Linux and were still not updating it?
Because some of us like to use our phones as computers, not just web terminals. If that's not you, more power to you. Some of the things I like my phone to do require it to work offline on local files, and the app model manages that pretty well.
But they could take Linux and do a s:linux:harmonyos:g on it. Not saying they necessarily would or did, but that's basically what they did when they announced the first Harmony OS to run on phones with Android. Sure, they called it version 2 instead of 10 and they had a different format for app packages, but it was Android in every detail. I'll believe they've made a completely new kernel when I see the images and the technical reviews of its contents, not before. I've seen too many people boast that they've invented a completely new OS, UI, browser, whatever which actually turns out to be someone else's whatever with a really minor change on it which breaks maintainability but doesn't do much else.
The camera's output kind of has to be the attack vector. You can poison the model in training, but it won't do anything useful unless you can get it to fail during use. If you can poison the training data to identify some specific input as something it's not, it won't help you unless you can also get the camera to properly recognize that thing in production. So, for example, if I feed in images of stop signs with pixels changed and convince the model to identify them as detour signs, but I can't get the camera to identify tampered stop signs in reality, then I haven't obtained any result. If, instead, I manage to get it to frequently mistake stop signs for detours, then it won't work during testing and won't get to production. Thus, to make malicious use of this, I need to be able to poison it in a specific direction and later invoke the behavior.
Image descriptions probably not too much because the processed images will be direct from camera ones. Since they don't contain poisoned pixels, they won't trigger the inaccuracies as often. While it's possible that an image of a purse gets mistaken for a cow, it's a significant enough change that the user will probably know about it. For generated images, it probably would cause some problems, but I'm unaware of a reason for a visually impaired person to generate images that's any different than a sighted person doing so.
From their download page:
"After the initial download, the app will download additional ML libraries and resources that will require stable Internet access and approx. 4GB of storage."
So my guess is that they download more stuff at runtime for Mac OS and include more stuff in the installer for Windows, possibly because it aids with signing for binaries. Either way, ML stuff tends to be pretty big. I suppose they assume that anyone with enough processing to run it probably doesn't mind the disk requirement. With the tendency championed by Apple, but unfortunately not limited to them, of including 256 or 512 GB of storage and not letting it be expanded in many of their computers, I would find such large tools constraining if I was running it on a typical hobbyist artist's computer.
Both cases have the much harder task of getting the system to react wrong when it takes the picture. IN this case, the pixels in your file are the pixels the model ingests, but in the case of a camera, it captures with some inaccuracy and lots of angle and background choices, so you have no guarantee of it getting whatever interference you try to send at it. Not that doing something to mess with either system isn't possible, but it will be more difficult.
"I agree, but skill without passion is not talent."
Why not? If your building needs running water, do you refuse to accept a plumber who doesn't plumb for fun? If you need something driven somewhere, do you only hire racecar drivers, who clearly love driving, rather than someone who can do the task correctly and does so just to get paid?
I admit that there is a correlation between the people who do the best work and those who have a passion for it, primarily because their passion means they've gained a lot of experience that people who don't have the passion don't tend to have, but there are two provisos to that. The first one is that this is not always the case. Programming tends to have more little details that take a lot of time to learn, and passion helps with that. There are other fields where it doesn't take so much passion to learn the required things to do the job well. The second is that someone is still capable, even in a detail-rich field like programming, to learn how to do something well without having a passion for it. If someone just writes code to get paid for it, but they spent the time needed to know how to write that code well, then they are as capable of making something good as someone who learned that because they enjoyed doing so. I've seen a lot of people who chose tech because they wanted something well-paid, not because they enjoyed it. Many of those people were bad at it and left because the hill of knowledge you have to climb was just too high. However, some of them didn't have any problem climbing it, go to work as programmers and produce excellent output, then go home and do something unrelated with the rest of their time. They are no worse than you or I.
This sounds like a change made without a spec at all. The fact that the local end got one day of work before the external provider started work suggests that someone came up with the idea and expected that everyone involved could begin work right now. I've worked for plenty of people who prefer to skip the planning stages because they don't understand why I would need any time. After all, they've just told me what they want, so why should I waste time planning when they've already done the planning? It often takes a lot of explaining to get them to understand that their idea is not a complete plan for its implementation, or more likely just that I'm not going to deliver what they want until I've done whatever it is I'm going to do.
Generally it's someone who doesn't understand technology trying to either prevent damage or sincerely being deluded. Some people really don't understand what is an attack and what is a bug report. I'd like them not to exist because it should be obvious, but unfortunately, some people just don't get it. The other category is trying to avoid having to do any expensive work to redesign the system. They think that the bug is so obscure that nobody else will find it, and surely nobody has used it yet, so if they can get this reporter to shut up, then they don't have to fix it and don't suffer any damage. As they see it, if they fix the problem, they'll have spent time and money they don't want to and the reporter will publish the report because it's safe, and if they do nothing then the reporter will publish and they'll get the reputational damage, but if they can muzzle the reporter then they can stop. A threat is their attempt to keep things quiet, often keeping them from having to fix their problem as well.
When they try it and it doesn't work, they find themselves in a vindictive mood and follow through on their threats to make the reporter pay. It's happened that way numerous times and will continue to for a long time. There's a reason why I tend to be worried whenever I find a vulnerability, because I've experienced this before. I've never been sued or prosecuted, but I have gotten some unhappy notes with low-end threats on occasion. The one pattern I've identified is that, if they responded that way and I didn't push back, the bug is still there today.
And if enough people figured that out, the admins would notice and stop you from doing it. Our quotas were just on the size of the home directory. I wouldn't be surprised that something along the following lines occurred:
Helpdesk: We found twenty students who were each using close to the 1 TB cap for personal files.
Admin: 20 TB. Wait a minute, that's 20% of the cap before we start paying more. For 0.05% of students. What would happen if 10% of them did that? [calculations] Uh-oh.
It doesn't, but fortunately, you can't bypass any controls you want with the cloud. The admins can put their important data storage on cloud-managed disks, their own managed disks, or a combination thereof and have the same controls or lack of controls.
And we ever had that working like clockwork? We moved from "the research data is in those boxes in the corner of the office, or maybe the ones that got moved to a storage shed when the corner was filling up" to "the files are probably on one of these servers that everyone has an account on but there's no organized backup" to "they could be on the servers with backups but maybe the researchers didn't upload them because they stored them on their personal computers" to "they could be in one of the university-provided data storage systems but people have been using OneDrive instead". It's not great now, but it's not like we ever had perfection before.
Archiving is hard, and there are relatively few university archivists for the stuff being generated right now.
I don't care what warnings they put around it. If they have a bot intended to mimic someone, somebody will eventually get that bot to print something they don't want it to. At best, that's an embarrassing video which gets shared and causes people to laugh about the AI company. More likely, those people think the candidate being mimicked asked for that and will mock them as well. At worst, some people actually think the statement actually reflects that person's views, which probably won't affect most viewers, but it doesn't have to. If I were the candidate here, I'd be very happy to see this thing taken down, and if I really had no contact with the place that set it up, I'd probably have been the one to send a notice to OpenAI to check for ToS violations.
The incremental cost involved in delivering something to me as well as my neighbor from the same starting point is less than the incremental cost of flying separate drone trips to each of us. Most of the journey can be split between the two recipients, and if many of my neighbors are also getting deliveries at the same time, it goes down much faster. The drone, meanwhile, has to keep returning to the base because it is not large enough to carry items from multiple orders.
The aerial aspect doesn't help either. You need to expend a lot of energy to lift things off the ground and keep them up there which is not expended when driving them along the ground. There are some savings with a drone flight, but the energy requirements of flight are not minimal.
I think that, no matter where they were, the blame could go anywhere the internal politics wanted it to go. It wouldn't be hard for someone to decide to blame either or both of those people. A charitable manager could probably have insulated both from too much blame and reduced it from firing to an unpleasant meeting. It comes down to the question of how high on the organizational chart did it go, how angry was that highest person, and did they want to do something to either of the people involved.
If you read past the first sentence, it would become clear that I was referring to experience with server hardware and racking to know things like the weight of the equipment and the general stability of racks. I have a feeling you did read past that sentence. Thus, your reply fails to make any meaningful point.
Throughout childhood, you also learn that it is not as simple as "heavy thing in high place, always bad". For example, a heavy item placed on a countertop which is elevated usually doesn't cause the wall to topple over because the counter is attached to the wall and floor. A heavy item placed on a sturdy and stable table is also usually fine because the table helps to spread the weight. That's even true if the heavy item is not centered on the table. Of course, there are numerous examples where it would cause a problem. Life would show you that there are cases where that works and ones where it doesn't, but unless you saw server racks in your childhood, it wouldn't tell you which one they would be.
A person who assumed, incorrectly as it happens, that server racks, being designed to hold really heavy things, would be designed more like a counter than a cheap bookshelf in the weight management department is really not that outlandish. If they went the other way, I could easily imagine someone not wanting to install a server in the high part of a rack because that's kind of heavy too, so maybe we should reserve the top slots for if we ever get some lightweight fan array or something. I am asking you to consider what things would be like if you were missing some important information that you have gained through experience. There was a time in your life when you didn't have that experience. We don't know whether this particular person was a new starter who had never worked with a rack before or just clueless, and there's a chance that the correct answer is both. There is also a chance that the answer is only one, and it's less stupid than you make it sound.
Rewind to your first experience and you might not know that it would make the rack so top-heavy and that the rack couldn't handle it. Sure, having a heavy thing at the top is a problem, but not if the thing in the bottom is also heavy. Those big servers are down there. I've never lifted one of them, but I did see someone go get another person when they were going to move one, so they must be heavy as well. This rack is built to hold lots of really heavy things, and it's not even resting those heavy things on the floor or on each other, meaning it was clearly built to take a lot of weight. They built it that high, so presumably they have built it to take weight at the top and it was installed to deal with that, for example being well anchored in the floor or wall, the way you would anchor a shelf if you wanted to put something heavy on it.
We know that those things are not true, but if you've just started on plugging in servers from working with desktops as a student, would you know all these things? I didn't know the typical weight of rack servers until I lifted my first one; there had been no reason for me to look it up.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
