Posts by david1024 225 publicly visible posts • joined 14 Apr 2015
The future is under the sea. Microsoft's underwater data center blazed the path, and now China is going to try it out with wind power. Space is really for sensors and communication (and maybe interceptors eventually). Here's a competing platform article on the Chinese effort: https://www.tomshardware.com/tech-industry/cnina-deploys-wind-powered-underwater-data-center
When the folks get angry enough a successor will arise. Same as GM/Chrysler/Ford created Honda/Toyota/Nissan and they on turn created Hyundai/Kia.
For computers, IBM created MS, and we are waiting to see what's next. The shear volume of money and systems, I think, hampers movements these days vs the 80's/90's.
Windows needs to be called xonfuse-ows. Where did it actually copy your file? And will it be there later? Was it reliably copied? How would you know as they hijack content to OneDrive and even command line tools are hoodwinked.
A prime feature of any OS is the ability to copy files. It is broken by design in win10/11.
But if all you want is something to watch cat videos and influencers with, I suppose win10/11 are a step up from android/iOS in that regard.
The solution to this 'problem' of non-human generated music is digital signatures...
The same as it is for photographs and images..... The same as it is for spam texts and calls and emails. Let them create the slop and let the market decide how long to finance it. The signature will let me pick AI music that I like or deselect and just listen to humans... Or a blend.
Pki solves all this with curated, trusted entities signing/attesting to the genuine-ness of the source of the material we see/hear. Not perfect, but it provides authentication for trusted sources. All we need is a red/green indicator or a hollow/filled dot for authentication status. Right now, I have no choice, I get a bucket dumped on my head and have to filter with no tools.
The key here, pun intended, is that everything is signed/marked--not just the modified or machine/AI content. This only works if _everything_ is marked.
Space is hard. NASA says so, it is continuously re-proved, and humans remain fragile.
While China has a good backup/contingency plan in place and seems to be using it effectively--it is always a good thought exercise to explore what would it take to rescue with non-Chinese (US/commercial in this case) assets.
How many of these passwords were for throwaway accounts that users were not routinely using?
I know I don't put much thought or care into a throwaway account password, and I bet bots and click farms don't either. And I have way more of those throwaway accounts laying in my digital wake than ones I really use... So how reliable is this approach of using compromised accounts? Still fun, but I don't think the data tells the same story today it would have a decade or two ago.
Noone thought that maybe a skip from 7 to 11 might work out since they'd had the 7 machines so long out of support? 300+ mill replaces a lot of hardware... That's going to need to be re-replaced now for an inflation adjusted swag of 450+mill. I can install 7 today if I needed to, not like it is illegal. Get waivers and pocket routers to give that old OS some protection and accept the risk as you go to win11 (or Linux or whatever)
That's the head that should roll for this.
Yes, and this was the problem Xerox faced, losing the name. Regulation and lots of litigation by Xerox let them keep the name and educate the public, but the Internet is much larger than the newspaper/magazine pool Xerox was swimming in during the 70's and 80's.
TBH, why even call it teleconference or videoconference?
What we actually do has nearly nothing in common with the activity, hardware, and infrastructure used when the terms were created. We share screens, remotely control virtual whiteboards and even computers, and have text chat all during these events that use digital IP pathways.
Videoconference is not what we are doing anymore. It is zooming or skyping.
Same with the text messaging... We send audio and video and shortcuts to launch zoom or Skype. There's end to end encryption options and groups with voting. And sometimes you need to install or update a client software to do it..
so, English, not having a word, steals the name and turns that into an adverb/verb.
Excellent video from ifixit. This is why I buy from them whenever I can even if the price isn't the lowest, they pay back triple the difference with information and a trustworthy supply chain.
One: who carries keys anymore and this device? The phone _is_ your key, unless you aren't flexing hard enough. The proles (like me) that still have keys can get a case.
Two: the increases screw count and internal rigid assemblies add stiffness you don't get from glue. While there are drawbacks, it is much stronger and easier to work on than glued construction. We've all gotten good at the glue, but screws are superior in every respect except cost/labor.
The only comment I have on the design specifically is to notice the big F.U. from apple on the spaghetti mess they made of the USB design. There's no reason to integrate the connection into the flex like that. (Especially with the obvious sale on internal connectors that was obviously going on during the rest of the design)
Supposing everything in the article is true, Amazon is training the next set of 'old heads' on their customers' dime... They have offloaded that expense. They will lose negligible numbers of customers. This is the benefit of being the leader/default option, and they are taking that money. And you'll pay--same as me while they train folks and when they get expensive, it'll be time for another round. :(
Lots of folks use notepad. Even if you just strip formatting with it and never compose with it--it is super handy. None of the things I use notepad for would be enhanced by AI, unless you let me dial in how that paste from excel looks. Of course, they likely want the AI's telemetry and don't care if I actually use that AI.... MS needs to use it, not us.
I get the vision, but even at the lower price points before broadcom's acquisition, the value wasn't there. Now only 10% of the customers that were forced into 2 years of shelfware acquisitions are onboard with them and even more lock-in.
However, what we may be seeing is that their previous agreements for those newly forced VMware features overlap with existing agreements.
And finally, at their licensing prices, pizza box servers and switchgear start looking attractive for onprem. $25k buys a lot of hardware these days. (And to get the most out of that license fee, you'd need at least a $50k+ server and a duplicate backup anyways)
ODC only makes sense if you are also in orbit to use the data and supply the inputs there. Otherwise Microsoft's under water DC was a much better idea. Much cheaper house calls and you can reject as much heat as you can generate (nuclear ships/subs reject 1000's of MW across the various fleets) and we are good at making those kinds of heat exchangers really cheap and reliable.
The real issue with ODCs isn't all the space weather, micro meteoroids, space junk, and cyber threats.... Or what happens when N.K. pops a nuke in LEO for testing... It is power generation and how to get rid of all that heat! Space is a terrible heat sink and power sources are heavy and dangerous.
The editing body could resign the image as they are the ones attesting to he fidelity of it. And if you trust, say the WSJ, then their signature would be good enough... But if I posted one, you'd want the camera's signature and mine on top of that. There are also ways to embed the full image and have the cropped one visible in the document... Would just be features to handle authenticated content. Just need to get it integrated so folks can understand it... Because right now, I'd say less than 5% would even understand how to interpret a signature in a picture... Need a lock icon or something...
Obviously the customer is wrong and needs to run things in broadcom's model. Further, once you really look at it, after some re-education, you'll be able to see how much better off you are paying more and be happy. In other news the chocolate ration is being increased to 8oz from 12oz.
Computers and networking are hard. You have to compete with all manner of industries to attract the talent you need too. Buy the computer, buy service agreements, pay folks to maintain them, and hope it holds together. And all the software license money leaves your local economy while the folks that actually do the work are stuck learning the new way we aren't changing the computer systems or watching cat videos while backdoors in the browser are pushing malware.
TBH, a competent admin assistant team with a floor of filing cabinets is starting to look attractive from an ROI standpoint. You got them locked in from a talent perspective and you can traina new helper in a week or so... No complicated industry certs, and you can lay them off if business turns down. Also, you aren't pumping money to wherever VMware lives.
TBH, we had plenty of big hospitals before computers took over. These systems are designed for accountants and lawyers, not practitioners and patients.
What boggles my mind is that such incompetence is allowed to persist like this. There is literally no need for these systems to be exposed to the internet. But, the lawyers and accountants aren't perturbed yet, I suppose.
I have always been a fan of KDE, and CDE. On Debian, they are what I expect in a PC. However, they have gotten a bit crazy in the past few years. Some of that seems to bring new devs/users, but isn't really my thing.
KDE is really what the folks want, IMHO, and there's a positive way to say all this without calling anyone's machine obsolete. TBH, the machines are not obsolete and "KDE+Linux will show you that your machine remains quite usable and even faster than it was under windows as you concentrate the resources on what you want to do." Easy peasy word smithing. Besides, they are competing with new machines, not M$
To me, this shows how to deploy AI more effectively. And that this study was designed to be easy for AI.
For now, Ai is bottoms-up and humans are top-down. Two separate flows.
I would guess that is because AI still isn't very good at finding the problem to solve and that's at the top/structual, and in that respect it is like a super technically competent green fresh developer. No sense, but knows how to iterate. AI doesn't seem able to mature, and the humans end up fixing code for all but the most straightforward assignments.
Actually. It'll react to that change as it occurs based on updated training data. This is one use of ML where it, the machine, outputs don't impact the inputs. (As opposed to something like a law enforcement use case would).
Expert systems would need new rules for the new climate situation unless they had that folded in already and that's really hard to impossible. Whereas the ML version of the weather predictor will work more like a me when I look at the current front/condition maps for my area and agree or disagree with the models based on my experience of real world historical data. (I'm usually more accurate than the models in use today for a day or two in the future, but not 100% of the time)
They are required to offer service, and the customers have to pay to maintain it. There should have been little info of value to copy.
There is no reason for the power company to have SSN, DL, or DoB (and a lot of that other data). Autopay should be handled by the payment service/bank, not the power company directly. This is bad operations. They should be concentrating on keeping the service secure where they should be the experts and leave the PII to the financial institutions (who should be the experts). This everything, everywhere, all the time info model is inherently insecure.
At some point, the go fast/break things folks...Maybe Don't understand what they are breaking. I think they know... They keep blaming sensors, but they need margins to account for that--and fix the sensors. Not just change the sensor and hope.
So knowing and understanding don't seem to be meeting up in the design from out here where I can see.
However, I do know:
Fast, cheap, right.... Pick 2.
Been an engineering rule forever. They need to push the triangle to the right a bit more. Maybe they have, wish them luck with me.
It's only a breakthrough as long as what it ingests is not AI and .... Is accurate/true. AI trained on general web content will happily spout out urban myths and incorrect technical information as long as it is reinforced in the sample inputs.
Llms set knowledge back at least 5 years and even then, it is popular knowledge, not necessarily fact. Plus, nothing original can be created. So, more like what a drunk teen would say at a party if they had really good drunk-memory. It conveys accuracy when it doesn't actually provide it. Like a teen.
This is easy to fix, authentication. Author infuses too much AI or other garbage... according to authenticated peers sitting on review boards... They get downgraded. The bonafides only get issued after in-person board approvals. Maybe farm out to university or have a funded professional body.
Pure AI and papers from other unrecognized sources should be subject to increased scrutiny.
Won't be perfect, but better than we have now. The publication system was supposed to prevent this, but it has not scaled to large input sizes, so we need to stem the flow a bit to get the need for reviews more in-line with reviewer capacity.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
