I can understand the sentiment, but in practice it's utopia. Sure the word processor may be 'complete' if you your needs don't change at all but that would often apply to software used in total isolation and in a lacked environment. In practice, the world changes all the time and some of those changes are kind of 'mandatory'. For instance, things like internationalization, unicode, UTF-8, bidirectional support etc. were not supported by early word processors but are quite needed in a global world. The spell and grammar checker (if such existed) could likely also need some updates as language evolves. New font and image formats have come out. These are things that will be critical for a lot of people and you would expect a word processor to keep up to date with this. Bugs may have been found in the document parser, making it dangerous to open untrusted documents. Most would like those fixed. Likewise, the hardware and software around it changes so Wordstar would need to adopt as well so you can run it. Then you could argue that Wordstar ought to keep adding such 'critical' updates to version 4 without changing anything else. But what if other users are craving for new features. Should Wordstar then maintain separate source trees for every major version and backport all those complicated 'mandatory' changes I mentioned? It's hardly feasible.

This lady got a 113K bill but managed to get it down to so far 31K. Still outrageous ofc. Esp because the rest was covered by the insurance so someone is paying nonetheless. https://www.vox.com/policy-and...

Very interesting story. I haven't heard of that early versions of Windows having .zip file support but you also say it was removed again. According to WIkipedia it was only in '98 (via a Plus!) package that Windows gained ability to handle .zip files. You refer to the 80s... File Manager (which would be an obvious place to have .zip support before there was a desktop / explorer interface) was introduced in Windows 3.0 in 1990. Is that where the functionality was?

As for your first comment, I can't prove a negative. I can just tell that in my experience true tech companies (IBM, Apple. Google, Microsoft etc.) don't tend to use scrums. Perhaps some teams internally use it but it's not the majority and it's not a company policy and none of their major products or successes were built using scrum. Whatsapp was sold for billions, were 50 employees and didn't use anything like scrum. Spotify is the one everyone refers to to praise scrum. Will they have over 5000 employees. Of course they are not all devs but it doesn't come across as a clear success story of efficiency that you can do a music service of 5-6000 employees. Regarding management control - ofc management wants control. But my point is that scrum merely makes them feel in control or that the information they get is indicative of increased progress, which is rarely the case. Regarding your point of productivity drop - do you deny the possibility of measuring productivity? I'm speaking from experience and the productivity drop seen through introduction of scrum (and others). Regarding the last part (brute force) - no, sw development doesn't need to be brute force. You can have teams (big or small) with tech leads who deeply understand what needs to be built and is able to create tasks himself, to delegate areas to others, to arrange proper touch points etc. and basically MANAGE the work in the way that neither scrum nor most project managers (or managers, for that matter) manage the work. Ofc you're dependent on a truly competent person who both understands the work and is able to manage it. Scrum is brute force because you think if you follow some magic / cargo cult process completely decoupled from the work and just push it hard enough, somehow magically results will come.

I just used those as examples for complicated software that would have been impossible to build using scrum. I could have used web browser, office suites as other examples. None of that has been built using scrum. Scrum is applied by mediocre companies who struggle to build even simple things. They think it's an elixir so they can become more productive like all their peer companies out there has supposedly become by using scrum. In fact they become even less productive. Scrum may work fine in super simple setups - think old legacy bank systems and where routine enhancement requests come in ("send out a letter to all customers in state X, Y, Z"). There of course at least you have a chance of defining and estimating the task and going through the rituals in any meaningful sense. It won't necessarily make you more productive but at least it won't kill your productivity and if it takes the place of other means of tracking progress etc. it may not even be a negative. But if you're about to build a word processor (like Word etc.), by all means don't hire 50 devs and start up the project using SAFe, scrum etc. It will be a money pit. :)

I'm talking about Scrum as it is done in practice. But it's very common to see people defend scrum by saying "Oh that just means you're not really doing scrum".

No team of highly competent developers in a true tech firm would use scrum. Do you think compilers, operating system kernels, databases etc. that we all use have been built by scrum. Of course not, it would take a century. No where scrum comes in, is in pseudo-tech firms. They can't recruit top talent and thus have a lot of mediocre developers who have to be told what to do. So ofc they have low productivity to start. Management would like to feel in control and see things delivered. Also, they tend to have a lot of administrative personnel with little to any technical insight. Management would like to put them into various roles to "drive" the project, but of course with very little insight into anything remotely related to the work. So they are put in as PM's, scrum masters and other tasks and what they deliver is pure administration / messaging passing, as could have been implemented by a python script. Then you apply scrum, hire more people than you ever were (both devs and admins), see productivity drop even further to the point that you have never delivered so little product per $. You still have tons of delays but you have more people to "explain" it and deflect the blame (often to outside) or find excuses. Eventually some half competent developer/architect will become the de facto PM (but without any credit, of course) if anything is to be delivered. Because there's so much reporting, management feels in control. Often it happens on project billed by the hour and on time-material basis so there's little insentive to optimize. It's basically a brute force approach to software development.

Yes that is often what people say to deflect any criticism of Scrum. In that way, Scrum can never fail because you can always say you applied it wrong ;) But at the end of the day you must look at what it brings to the table when applied as is done in 90% of places. And there it's a cancer. It's a mind virus. But it has been here so long, many cannot even imagine how productive it could be without. "Everybody's doing it so do it too..."

But I have heard many other people say this. If he used chatgpt himself for just a few hours and used it imaginatively (such as combining several subject, concepts and genres into one query), he would realize it can't possibly work through stitching together text. The combinatorial explosion is simply too big that there is something to stitch together. It shows it really does operate at a more conceptual level with some level of understanding. Even more so for gpt4. I wonder if it is a kind of denial. People see these responses and their first reaction is that it appears too good to be true. They instictively always thought AI would not be possible in their lifetimes so once it is there it must be fake somehow... And the only way to explain away that it produces coherent responses is to claim it is stiching. Maybe I had the same thought when using gpt the first few hours. But as a good scientist I quickly had to abandon such a hypothesis since it clearly wasn't consistent with the data.

Yes it's curious but I don't think it is just an issue of increased power demand by software. The normal battery meter doesn't work by calculating /estimatingin software how much power was used, this would be too unreliable. Rather, it uses a hardware coulomb counter to actually measure the amount of current leaving the battery. So even if there was a new software telemetry feature using power, that power would automatically be accounted for in terms of battery actual vs design capacity. If it's a software error it would be in the algorithms that collect the current information and does the calculations. But it works on other phones and if it was just that I guess Apple would have fixed it (it's bad PR). But it would be interesting to know if the drop reported corresponds to actual user experience. 20% drop over less than a year is way more than expected.

ChatGPT is just showing search results?! Have you even tried it yourself - especially gpt-4 but also just the free version (3.5) for that matter? Then you would know what it does cannot be explained just as showing search result. Ask it some question that combines enough elements/subjects that it is extremely unlikely anyone would have asked this (even in different wording) and you can see it synthesize a new answer that shows understanding of the subjects... it is not just copy/past-ing and adjusting/reconfiguring. Of course everything it knows is in the training data but your description of how it works is no more true for gpt than it is for any human being.

In 2013 I thought it would be cool to work on fusion etc so I visited their web page. The project had already been running for a long time. It seemed completely fake. Everything was extremely superficial and old and there was no information on status etc. It was basically like the PowerPoint content you imagine would be before they started. There was more about how the project was administered than about the actual results. They didnt have a list of "Careers" or profiles sought - instead there was a page explaining that if one wished to be hired one needed to contact a national agency (i dont remember it was a Ministry or university) in ones country of residence to hear of opportunities and employment procedure. It was all so opaque and just screamed of bureaucracy and nepotism. Only a few years later I read the first article critical of the project for being inefficient and for lack of progress. This one is just the most recent. Probably the best course of action is to completely demolish the project and start over. Avoid the temptation to try and use and work products, materials or human resources from the previous project. Even if it can seem like a good idea there is a strong tendency that such get tainted by poor culture at the old project. I wouldnt even be surprised if there arent that many things produced at all. Have they even built tje building that would be the reactor hall? Or is that also only in "planning"?

Didn't even know it was a thing, anywhere. In Denmark the last remnant of that system was shut down 10 years ago but in practice it was much longer since ordinary people could even get a check book. I'm 41 and never had a check book or wrote a check. I've gotten and cashed maybe 3 checks in my whole life, the last one being in 1998 or so ;-) even before mainstream internet (80s, early 90s) people used electronic banking here... First touch phone banking, later via PC dialing in directly to the bank via modem for free. Direct transfers (also interbank) has been possible in ATMs as well since early 90s. It's been 11 years since i was even in a bank and that was for a mortgage. Probably 20 years since the visit before that. Even the mortgage can now be done completely remotely.

Refernces are not the same as smart pointers. And true, smart pointers don't selve everything. When the lifetime of the resource is complicated and can't be tied cleanly to lexical scopes then no amount of language support can safe you, not in rust either

The Linux kernel is written in C. You basically have to deal with pointers and memory in very low-level ways all the time C doesn't have the safe subset that C++ has. C++ (like rust) would be beneficial since it still allows you direvt access to memory when needed (to access I/O registers etc which usually isn't where the problems occur) and you can still write everything else using safe constructs