Based on the Iran war situation I don't think we should be building more datacentres for security. They are easy targets. We should be concentrating on resilience and that means distributing capacity and capability where possible.
You’re describing having a bunch of data centers in different location and enough redundancy. What makes you against data centers themselves? They are just a way to pool resources to benefit from economy of scale. They don’t have to be enormous
Centralised "economies of scale" mean consolidating risks into geographical and corporate ownership. I mean look at the current situation: that consolidation means there are a few corporate players, any of whom could just pull the plug on a huge amount of infrastructure in a war or other geopolitical mess.
Also we have a layer of abstraction above the datacentre now which is the cloud provider. And that does not necessarily (especially in our case) have an economic advantage. And it is again a single point of failure. One cloud provider compromise and the scope of compromise is across multiple datacentres and businesses and potentially national governments.
I'm suggesting bringing a lot of stuff back in house or within tens of thousands of small datacentres where there's a few racks max. And we keep our abstraction depth low.
I'd go as far as designing things to be permanently disconnected or just occasionally connected these days. Even single-user stuff reaches into clouds and datacentres when it doesn't need to.
Yes it would still. Which is why I think the process is misguided. I mean look at Hungary which was a near miss. It needs to be resilient to state failure as well.
This means that the entire idea of a corporate EU spanning hyperscaler should never exist.
I look forward to the day we have a sovereign CPU, RAM, storage, ancillary ICs, production line, supply chain, software stack and associated infra than I can walk into a shop and buy and use myself.
I don't think it'll ever happen though. These initiatives are mostly fluff. Throw everything into AI because it's the current fad but not even look at stuff that runs everything RIGHT NOW.
If anything was to happen war-wise, we'll be running everything on recycled trash.
>If anything was to happen war-wise, we'll be running everything on recycled trash.
ESMC should be online in a few years with 16nm class ICs. That will be tech that's over a decade old at that point but it's also "good enough" for anything except AI training.
We have now sovereign CPUs but we are struggling to get HPC ones. They will be there soon though in 5-6 years. RAM is already possible for Europe to manufacture if they can get the fab. Supply chain remains a question mark for the raw materials. Software stack is almost there.
In effect, I would say we are around 60% there. The most important thing actually missing is Fabs. Everything else I see a straightforward path with money and time.
There is a big structural problem in Europe with these sort of initiatives: people receive grants for attacking identified problems; if you actually solve the problem then you have to find a new problem for your next grant, so it's best to not actually solve the problem but claim your next proposal will.
Combine that by not actually wanting to reward people properly when problems are solved and you have the root of why eurotech has totally stagnated.
I would turn it differently. The grants are awarded but they come with requirements that are impossible to satisfy without severely degrading the quality of the product.
1. Most grants are one-off before the EU moves onto new fields of grants. Therefore, there is no way to build a deep expertise in a field for one particular center.
2. Most procurement are very top-down 1990s manner. This is no way to build infrastructure (by mandate). You need to fund competing proposals and then select good things from each of them. Instead, EU has this propensity to award only one single grant, which means every single big org that will be affected by this grant is on the proposal (probably), which then means that there is no room for opinionated stuff.
3. Most EuroHPC funding is national co-funding based, which means same proposal needs to be submitted 1 + x times where x is the number of countries in the proposal. Due to requirements of EuroHPC, x > 3. So, 4 proposals, 4 negotiation for any single idea. It is such a mess.
This in addition to bureaucracy where there is no dynamism in contract negotiations. There is no flexibility in rules. The EuroHPC and every single govt would be okay with projects failing rather than allowing for a small monetary rule change which allows the project to succeed. (This is not even about getting more money but transferring money from one bucket to another in the same project).
I can go on, but these are structural problems why HPC in Europe stagnates every day.
No true Scostsman, aye me lad. Now really, because big bang doesn't work, means nothing else should be even tried? Are you even from the EU to actually know, or just feel bothered by the anti-US sentiment oozing from those initiatives? If that helps, my feeling is that it's not anti-US, just a normal reaction to the acts and thoughts of the beloved best-leader-ever ruling the US right now, and his faithful elite. It's trying to protect oneselves, maybe a bit of rallying under the (blue) flag, and defeatism has no place in it.
I'm in the UK. My partner is German. I spend a lot of time in the EU.
It's really the "this is fine dog" realising that it's not fine and apathy and poor governmental and corporate decisions over the last 30 years have blown up in our faces.
I'm not sure that the Grauniad has a particularly good global reputation for independent and critical journalism. It publishes the same mix of disguised opinion pieces and rather biased junk articles as the other side of the political spectrum.
There isn't a single news source that you can trust as such. You have to compile a lot of them, remove the unverified information and see what is left. Usually not a lot.
Whilst not commenting on that, a fascinating quote from the article you are replying to is:
"Viner also oversaw the breakup of The Guardian’s celebrated investigative team, whose muck-racking journalists were told to apply for other jobs outside of investigations."
This tells you something about why you might feel that way.
I mean they operate as a trust and wear their journalistic bias proudly on their sleeve; in terms of intent their altruism is self-evident.
That said, no British media is exempt from adherence to D Notices and tenets of their legal system like the concept of a super-injuction, whereby a court order prevents the reporting of the fact that the injunction exists at all.
That the term was coined by a Guardian journalist covering the 2006 Ivory Coast toxic waste dump scandal should be context enough as to their motives and constraints.
The reformed DSMA notice system which replaced the D notice system in 2015 is somewhat more specific on what should not be reported. I think that's fit for purpose now. And it's still not legally binding. It's an agreement. Thus it does not break press freedom should the notice be found unethical or covering something up.
I sort of disagree with this. I bet if you asked liberals and progressives in a country like America for a foreign newspaper they read -- if they do read one -- in most cases it is probably the Guardian. So it may be only the best of a bad bunch but it does have that reputation.
Never thought of it from that perspective. They should read multiple sources too. BBC as well, which once the articles have settled, are quite good. (just ignore the breaking stuff which is dubious sometimes)
It's an interesting fact. I have been surprised at how often The Guardian sneaks in to stuff that is otherwise overwhelmingly North American sources only. The Legal Eagle channel on YouTube has cited The Guardian a few times that I have seen, for example. And they even present it seemingly on the assumption that they don't have to explain to their viewership what The Guardian is. They just slap it on screen the same as they do the Washington Post or Politico or some such.
Don't know if it's just my settings (spoiler: I read the Guardian) or it's by tracking IP addresses, but at least for me, when I go to the Guardian, it defaults me to a US-specific home page. I assume British people see a different page, with more coverage of their government.
Well sure, of course they should. I'm just telling you how it is. All anecdotal, but in my experience, if you ask a liberal or progressive American for a foreign newspaper they sometimes read, at least four times in five, they will tell you the Guardian. They don't see that as "only getting their news from one source." They see the Guardian AS one of their "multiple sources," because at least it's "not from here."
Edited to caveat: of course, if you ask an American what foreign newspaper they read, most of the time we will say none. But if we do, it's probably the Guardian.
The BBC used to deride Sky for being "never wrong for long", but the race to "break the news" changed that. If the news is about something that happened today it's barely worth looking at.
Personally I get my general news from "The Week" magazine each week, which occupies half an hour on a Saturday morning. It has a selection of articles from across the UK and international press, cut down to give an idea. This week I see ones from The Observer, the Financial Times, The Sunday Times and the Spectator. There's a coverage of america, with input from the NYT, Washington Post, National Review, New Republic, Bulwark and Politico. Elsewhere coverage of Cuba includes stuff from Global and Mail in Toronto, Diaro de Cuba and El Salto in Madrid and 14YMedio in Havana.
The Guardians role in modern UK society is to launder right wing talking points through a few layers of progressive sounding rhetoric so that the average person on the street can say "Well if even The Guardian agrees, maybe there is something to it."
They have been trying to kill people in the UK for years. And have been funding proxies everywhere, some of whom have attacked the UK. We're not really even involved and I find it hard to agree with this point.
However it should have been dealt with earlier rather than latent bombing.
I do and I agree with him. Iran is just one of many on a long list of supporters of terrorism which includes the US, the UK and Israel, plus many more countries like Turkey and Russia.
I’m never hosting or dealing with any companies in Iceland. I had a run in with a hosting company there who was DoS attacking us from compromised nodes. I emailed them and they told me to get a letter from a local lawyer telling them to stop and they’ll look at it. In the end we contacted our DC provider and they dumped all traffic from their entire blocks.
A year later same attitude from a different one hosting a web site for Covid misinformation which was against their own AUP.
As a mathematician by trade I think they’re overblowing it. You can choose to use it or not. I choose not to because I enjoy the process. But I’m not doing formal research or getting paid to do it these days.
I will note that the average corporate mathematical modelling is usually a fucking circus so adding AI might make it better.
This is becoming less and less true unless you're specifically talking about usage of it outside of a work environment. Many work places are requiring people to use it and/or tracking usage. I don't know about in academic settings, but I'd imagine it's becoming heavily used there too?
The choice only remains if using it isn’t a huge multiplier. If it is a huge multiplier/accelerator, then for a while it will be ambiguous and the choice will remain. But as time goes on, the gains of using it will be so apparent and the advantage of the people who use it so great (in publication numbers, hiring, etc) that it will force others to.
I don’t say that with any particular relish. But I am skeptical of the choice angle past a certain point.
I don't think all universities or research agencies are particularly pressed on this. I mean my daughter is a notable researcher in a scientific field and they have absolutely no pressure to use AI to pump out papers or deliver value quickly.
I highly doubt there is any overt pressure in academia right now to use AI. It’s a relatively conservative institution. But there’s certainly pressure to publish (publish or perish being a common phrase for decades), and competition for jobs in academia is fierce. That’s what I meant in referring to long term pressure.
Well I rather like to be paid more than a mathematician so left academia rather quickly. In my case corporate modelling mostly involves making prediction models based on shitty data and metrics to make poorly contrived business decisions that lose millions of dollars.
reply