Posts by OhForF' 655 publicly visible posts • joined 29 Mar 2022
So the EU should not be allowed to restrict what foreign (or at least US) entities the companies in its jurisdiction may do business with while at the same time it is perfectly fine for the US to go after European companies when they do business with anyone on their sanction list including Karim Khan, Prosecutor of the ICC.
You need to be a pretty firm believer the US are the goods guys and can't ever do harm to buy into that argument.
Giving technicians based in Canada access to other regions allows them to support those regions during normal working hours in Canada. OVH did probably not set up separate teams and infrastructure for Europe, North America, Africa, Singapore but tries to balance both work and other load between the regions.
While they do mention "enhanced compliance" and "Improved data compliance" 1 for their Local Zones in Public Cloud they can probably not guarantee data sovereignity any more than Micros~1 can.
1 What does "enhanced" or "improved" compliance mean? You are either compliant or you are not.
The legislative power in the EU is the council of ministers. Nowadays the claim is that parliament shares that legislative power with the council on equal footing as parliament has to approve the laws in the normal process. Why the council should have the same (or in some cases more) power than the parliament that is directly elected by the people is still a valid question. I do not see a compelling reason why parliament needs the commission to put forward a proposal either.
The bigger problem for votes on changes to GDPR that benefit big companies is that big companies can afford more lobbyists than the ordinary people in the EU. Most members of the european parliament act in a way that suggests they are not aware their primary job is to represent the people and act in the peoples interest.
>What other option is there? <
Changing the way the "social channel" work, e.g. having the device joining an ip multcast group which should be transparent to the WiFi part of the net and doesn't require to drop out of the user assigned frequency to monitor channels selected by Apple.
Not sure if that would work for this specific service as i am using option 1 (don't use the service).
Your average user will not be able to figure out why less used channels aren't working properly for streaming and will not be aware of the option to fix it by disabling a service they might not be using consciously. Apple should at least communicate more proactively that this is happening when the service is running if even network researcher are surprised by it.
Too many SMEs have built critical workflows (usually billing an accounting related) on Excel spreadsheets so flimsy they do not work with a different Excel version. Updating the Excel magic to work with a different version is not possible because anyone that understood how that deep magic works has left long ago.
Micros~1 is only partially responsible for that.
The Ubuntu life cycle say 24.04 LTS (Noble Numbat) is in standard support until Apr 2029 so for now there is no argument that you can't just keep *using* RISCV unless you have some compelling reasons to upgrade.
There is money to be made if you can convince a CEO his attempt to replace employees by AI only backfired because he and the remaining employees didn't have the necessary AI expertise.That is already enough of a reason to tout you have the best AI experts available.
Additionally Accenture has to convince their customers that only they have the experts at prompt engineering you need or even the most retarded CEO will eventually figure out he can write that "give me an excuse to fire 10,000 people" prompt on his own (and save the money paid to Accenture).
I live in one of those countries and can confirm not everyone has an ID card and we are not required to carry any ID.
However any LEO deciding at his sole discretion that someone did something suspicious can ask for ID and if none is produced take the subject to the nearest guardhouse for identity verification.
Not sure if you folks in Blighty want that kind of "optional" ID.
On the other hand if you already carry a smartphone all the time the possibility of ID cards being used to build a location profile should not be your biggest concern.
Someone that can not distinguish Word and Excel is unlikely to have arrived at the IT manager position after rising through the ranks of IT. He probably wasn't hired because he can manage everything without having to know about the subject as according to the article he didn't have any people skills.
Without additional knowledge my assumption is he was one of those managers appointed for who they know not what they know or can do.
"There are no two people with the same workflow for deploying to production,"
"developers using Copilot to produce wonderful Terraform code, "but they don't understand the implications of their changes."
So the problem is vibe coding cowboys that do not understand what they are doing and can't be bothered to even follow a preset procedure for deploying to production.
The solution obviously is using more AI - can't start doing the sensible thing and hire developers that know what they are doing.
As Micros~1 set 34.000 full time engineers to work on the Secure Future Initiative back in Sep 2024 Senator Wyden will surely praise them for the great results that Micros~1 will publish any moment now.
Linux seems to have announced the merge window for 6.17 late in July and Kent Overstreet's pull request as linked in the article is dated Mon, 28 Jul 2025 11:14:33 -0400.
This article is not about Palmer Dabbelt and his late RISC-V patches (the article has a link to the el reg coverage of that as well) but talks about a separate issue.
What would be the output of that analysis, i.e. what do you expect the state machine or LLM to deliver to help with the system migration?
You can probably automatically translate Cobol to some other programming language like C(++) or Java or Rust but you'll end up with code that is harder to understand than the original Cobol and thus will be even harder to maintain or adapt to new requirements.
>Jake didn’t tell anyone about his fix.<
I wonder if the problems with the CT scanner started when someone got rid of a tripping hazard and decided to replace the direct cable connection across the room by properly connecting the CT scanner and the workstation to the existing wall connectors without telling anyone.
Do you really expect Trump will appoint a statistic chief that will always report low numbers so there is no chance they'll ever have to revise the numbers down? Every administration will want to publish higher job numbers rather than a lower number. As long as there is an error margin anyone appointed by the current administration will report the upper limit. As long as that is backed up by real data and not made up numbers it is expected to some degree and the number provides has some use.
If you have a closer look at what the US Bureau of Labor Statistic publishes you'll see that they currently provide an estimate for July 25 and list the numbers along with the 90% confidence interval and even show a nice chart where you can see the confidence interval graphically.
So e.g. for the total no farm jobs they say their current best guess is 73k and with a probability of 90% the real number is somewhere between -62.8k and +208k. I'd totally expect any politician talking about this number to say 'We had a big increase in job numbers, non farm jobs alone might have been increased by more than 200k' and that would not be a lie even if the reality might turn out to be a loss of jobs.
Anyone saying "they got their numbers wrong" when more data is available and the report is updated doesn't understand how statistic works and should be laughed at when they talk about it in public.
There is a "--dry-run" standard JVM command line option which creates the VM but doesn't execute the main method, documentation says it might be useful for validating the command-line options such as the module system configuration. Not very useful to test what a tool implemented in java would do - as you correctly point out that would have to be something the tool supports.
>If the efforts of the human and the AI are combined in an optimal matter<
Do you have a proposal how to do that? A bad combination may well end up with a higher error rate than either Eh or Ea.
A human modifying text prompts to get the AI to generate correct code is most likely not an optimal combination. The AI making code suggestions on what it observes the human doing (code completion on steroids) doesn't seem optimal either.
Even if you have a good combination it will not improve the combined error rate much if one of the "imperfect heads" is much worse than the other (e.g. an unskilled human attempting vibe coding combined with an AI where 1-0.95 error rate is optimistic but still better than the human that doesn't know how to write code himself).
>We are a support organisation providing support for a number of applicatins<
Your boss should be glad that AI isn't better or as good as you are on that job, If it were your clients would replace your organization by an AI. The customers might well do that before your boss makes back the money invested to replace you.
I'm really interested to see how technical experts in government "deploying and diffusing" emerging technologies will accelerate scientific discovery. For the life of me i can't see how sprinkling emerging google AI tech around government systems is going to help securing those systems either.
Icon for Peter Kyle -->
The authors of the papers are not idiots using AI, they are trying to stop lazy "reviewers" from wasting their time and tarnishing their reputation with AI generated reviews.
Not that i agree with that method - it would be way better to find a way to identify AI generated reviews and get the science community to agree that anyone trying to pass off AI generated reviews as their own work should take a big reputation hit and no longer considered to be worthy to do any reviews.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
