Updated to reflect our strong preference for US/CA. We have made exceptions in the past.
Since there are comments about this above, I want to highlight that we've hired successfully on HN for several roles for our current small team. We hired 1 of 2 roles last month, and are still looking for a candidate for this role, which is why I'm posting here again.
The main flaw of this article is comparing a general-purpose language built with production systems in mind (Python) with a domain-specific language designed for interactive analysis (R)... Beware of comparing apples and oranges, because productizing R code typically requires rewriting it in another language.
I've found even "normal" sized phones, like my S22, can encounter UI bugs where crappy interfaces appear to expect iPhone Max sized screens and so end up hiding bits of the interface behind button bars or fixed hight screens that hide the "next/submit" buttons behinds keyboards. I can only imagine how much worse these banking apps are on a smaller screen.
Sometimes they are just impossible to use. For example I could not hit the button to use Steam Guard in the Steam app to log in on one device. The screen was just too small for what the app expected.
They allocated €37.4 million [1]. As an European, I truly don’t understand why they keep ignoring that the money required for such projects is at least an order of magnitude more.
Deepseek's release has shown that there's no great risk in getting left behind. All the info is out there, people with skills are readily available, creating a model that will match whatever current model is considered frontier level is not that hard for an entity like the EU.
For everyone here shouting that the EU needs to do something, be a leader, what have they lost so far by choosing to lead in legislation instead of development?
They've lost nothing. They've gained a lot.
They can use the same frontier level open source model as everyone else, and meanwhile, they can stay on top of harmful uses like social or credit scoring.
Also speaking as a European, legislation is kind of the point of a government in the first place. I do think the EU goes too far in many cases. But I haven't seen anything that makes me think they're dealing with this particular hype train badly so far. Play the safe long game, let everyone else spend all the money, see what works, focus on legislation of potentially dangerous technology.
> legislation is kind of the point of a government in the first place
I would personally consider legislation to be but one means to an end, with the point of a (democratic) government actually being to ensure stability and prosperity for its citizens.
In that framework, "leading with legislation" doesn't make any sense—you can lead with results, but the legislation is not itself a result! Lead with development or lead with standard of living or lead with civil rights, but don't lead with legislation.
Your formulation sounds like politician's logic: "something must be done, this is something, therefore we must do it". Legislation as an end in itself. Very interesting.
> I would personally consider legislation to be but one means to an end, with the point of a (democratic) government actually being to ensure stability and prosperity for its citizens
You're correct, in retrospect I was a bit hyperbolic in my statement.
A better statement of my view is: the goal of a government should be the prosperity and wellbeing of it's citizens and the greater system we're all a part of (both geopolitical and ecological), and the best way we've so far discovered to do that is via legislation of an otherwise free market.
> They can use the same frontier level open source model as everyone else, and meanwhile, they can stay on top of harmful uses like social or credit scoring.
We are dependent on models created by USA and Chinese companies for access to the technology that seems to be the next internet - while the entire world is accelerating hard towards protectionism and tariff wars.
Yeah, this is exactly what scares me. But it also scares me that there's almost zero oversight on what USA and China are producing and the bias that could be embedded into these models by their creators...
I'm just not sure whether it's worse to be behind or to try to be in front by all means necessary.
I partially agree with you. The only problem is that these markets are highly monopolistic, and we will be creating another technological dependency on the US.
Deepseek didn't show anything except the compute cost of final model. We don't know how much data collection costed, how much unethical data like copyrighted data or OpenAI's data is needed, the cost of experiments etc.
> Creating a model that will match whatever current model is considered frontier level is not that hard for an entity like the EU.
If they have this as their top priority and allotted few billion dollars then sure. Not in the current form where the people involved are only involved for publication, not doing hard engineering things that takes months or years and they could do the same thing in OpenAI or Deepseek for like $1 million salary which both of them pay.
I’ll add some European wisdom to your sports metaphor. You don’t have to become a big football player to make money in football. I’d rather make money from the tickets and rights than dedicate my life to a sport that’s only played in the US.
> As an American, most of this post reads like doublespeak satire
Yeah, you guys have a lot of brainwashing to get over. I can imagine that you're deeply conditioned to read any outside views on politics as satire.
One kind of brainwashing is the need to reframe everything political into sports metaphors. The EU is not a sports team. It's a political entity. Whatever you might have been taught, these are very different things, with different needs. You can't have meaningful conversations about a political entity via sports metaphors.
Well, maybe in US politics you can. There you have two teams determined to beat the other at all costs. EU politics isn't like that. We are trying to work together, not kill each other.
> There you have two teams determined to beat the other at all costs
On the surface. It's all kayfabe though; heels and babyfaces. Just like with wrestling, the media know the score, and all the angles. After the match, they all laugh and joke together on the depraved billionaire owner's megayacht.
I don't think this is accurate, looking at it from outside. I mean, yeah, they both want to end up sitting on the billionaires yacht.
But one side wants to do that while looking out over a fascist dictatorship.
The other side has some weird idea that the billionaires will use their wealth to create a good life for everyone else too. Even though the term went out of fashion, it's still trickle down economics.
These two sides are not the same. They're both bad, but one is much worse. The last time fascism took hold it took nuclear bombs in Japan and firebombing Dresden to end it.
If you think Democrats were "determined" to beat Trump "at all costs" this last election, please, explain why they couldn't promise to stop arming Israel's genocide.
Democrat elites knew full well, as did the world, that 77% of Democrat voters wanted an arms embargo. Everyone who cared to look knew that in close battleground states over 3 in 10 Biden 2020 voters were saying that their vote could be affected by this, such was their feeling on this issue - understandable, given the daily atrocities livestreamed around the world.
Kamala's campaign had an easy win, a landslide victory for the taking. All they had to do was promise an arms embargo. Instead, she promised to keep sending bombs "no matter what"; even before her campaign page had a single policy on it. Does that look like they were fighting to beat the Republicans "at all costs"?
... How did Democrats hold Trump accountable for his insurrection attempt? Did that look like 'determination to win' to you? [0]
Did Blinken, Miller, Patel, Karine etc look or sound any less cartoonishly evil than the Trump goons? The things they said up there were mind-bogglingly cruel; staggeringly disrespectful of our intelligence.
The Biden admin censored millions of posts, pushed knowingly fake stories, physically removed journalists from the press room for asking legitimate questions, etc. Not super Democratic.
Sure; Republicans are much worse on some issues. But this corporate plutocracy didn't come out of nowhere. Despite the claims from the current Dem team, they do in fact take a lot of money from "bad" billionaires.
Even now, Democrats are crowing that Trump isn't deporting as many immigrants as they did. Because that's what's important right now??
Biden's admin pulled out all the stops to shut down student protesters - compare that to their response to Musk raiding the Treasury!
The two sides are not quite the same (neither are heels and heroes, ya know), but they are funded and owned by the same people. The difference is very clear when you see the unanimous support across the political and media class for things which the American people don't actually want - forever wars, environmental exploitation, tax cuts for the wealthy, full on genocide etc.
What do you call it when one of the billionaires who proudly and publicly sponsored the winning candidate for president (who, fun fact, has nominated enough supreme court judges that they ruled he has total immunity on official acts, and got away with an insurrection, rape, financial fraud, so obviously not a qualified candidate) does a Nazi salute on live TV in front of the whole country?
If this were a book or a movie, everyone would be dismissing it as far too obvious and not how real world works.
You’re right, it has become a fascist oligarchy in just two weeks. Faster than it took for you to realize it. But there’s still time to make Trump king.
> EU politics isn't like that. We are trying to work together, not kill each other.
Oh? Been quite a while longer since there was war inside the US than war inside Europe. While it's been no time at all since vicious party battles in major European countries. Or countries nope'ing out entirely. But apparently fascists are only a thing in the US now?
> creating a model that will match whatever current model is considered frontier level is not that hard for an entity like the EU
What industry has the EU caught up in or maintained pace in like that by "leading in legislation"?
I'd probably much rather retire in the EU than in the US but... there are certainly cons, not just pros, to the lack of urgency and bureaucratic "lets throw words at the problem" approach to economic development.
> Been quite a while longer since there was war inside the US than war inside inEurope
Maybe you need to look up why the EU exists in the first place. Way back when it was called the European Coal and Steel Community (ECSC). There haven't been any wars between member countries since it's foundation, so I think it's working pretty well, actually.
I don't have much skin in this pissing contest, but I've lived in Europe, North America and East Asia.
> What industry has the EU caught up in or maintained pace
What industries is the US leading which reflects itself in improvement of quality of life of its citizens? Cause some things really don't matter in grand scheme of things.
EU has better quality of life on borrowed time.
Yeah, some things don't matter in grand scheme of things. When your grand is larger, eu is in a bad place.
I’ve been hearing that about EU, China and Japan for the past 20 years in different degrees. Two generations were born and a lot of people have passed away in that period.
Even if EU’s QoL is on borrowed time, at least it has it for some time.
Personally I'm rather happy that the allocation was not too large at first, even that is quite a sizeable sum. The EU is great at kickstarting projects that sound like a panacea, but end up not leading to anything. Once they have something to show, by all means, throw more money at them.
The trap that these EU projects typically fall into is that they burn all of the grant funding on paying politically connected consultants to write reports. No one gets around to building an MVP.
As said before in another comment. The project can likely make use of 'free' EuroHPC resources, which will also be funded simultaneously with hundreds of millions. Still not Stargate, but if they can actually innovate something beyond the obvious (like R1) I think the money is still useful.
On what basis are you are stating this? I'm asking because I have been involved in another project like these (15M budget) and the main issue was the lack of computing resources allocation, because no one thought about it (true story).
The application process IMHO is quite complex (thwy want compute estimations and CVs etc).However if you figure how ist works it is at least relatively easy to get batches of 100k GPU hours via EuroHPC. Currently few calls are open, but there is typically at least also national infrastructure. Again this is nothing compared to what OpenAI or Meta has access to.
I just got 25k node hours on Meluxina for a fine-tune project. My colleague got quite some GPU compute oh Germany's Tier3 NHR Clsuter (Horeka with >700 A100/H100 GPUs).
Because America has the best private capital and startup ecosystem in the world it has a good chance of picking the big winners. There is no corresponding European ecosystem, only a bunch of small national ones. In fact, European investors are not betting on EU startups because they are unlikely to be able to scale to beat US and Chinese competitors due to lack of market and capital market scale.
Developers are cheaper but not a lot, and regulatory costs are higher - the difference in the end in not huge. Plus: if you want to win a global market you need global talent at global wages. You need a lot of private capital and Europe does not have any.
If you mean the NDICI stuff, that's hardly 'sending money overseas', and it's a fairly tiny fraction of spending.
> and to spend on illegal economic migrants.
... What are you talking about here? What portion of the EU budget is spent on that? What activity specifically?
In the real world, most EU spending is on regional development, agricultural stuff, and operating the EU (civil service, enforcement bodies, etc etc). The EU is not a country and has only a very small budget (about 170bn/year).
DeepSeek had plenty of R&D expertise which were not included in the (declared) model training cost. Here we are talking about building something nearly from scratch, even if there is an open source starting point you still need the infrastructure, expertise and people to make it work, which with that budget are going to be hard to secure. Moreover these projects take months and months to get approved, meaning that this one was conceived long before DeepSeek, thus highlighting the original disalignment between the goal and the budget. DeepSeek might have changed the scenario (I hope so) but it would be just a lucky ex-post event… not a conscious choice behind that budget.
Aleph Alpha is a business that has been going for some time in this sector, at least a couple of years with commercial LLM products. It's likely they'll provide hardware and base models for this project.
DeepSeek probably spent closer to two billion on hardware. And then there’s the energy cost of numerous runs, staff costs, all of that. The 5.5m cost was basically misleading info, maybe used strategically to create doubt in the US tech industry or for DeepSeek’s parent hedge fund to make money off shorts.
I mean, I get that the current strategy by most participants seems to be burning billions on models which are almost immediately obsoleted, but it's... unclear whether this is a _good_ strategy. _Especially_ after deepseek has just shown that there _are_ approaches other than just "throw infinite GPUs at it".
Like, insofar as any of this is useful, working on, say, more techniques for reducing cost feels a lot more valuable than cranking out yet another frontier model which will be superseded within months.
I am a hybrid profile, somewhere between a software engineer, a scientist, and an entrepreneur. At heart, I am a builder who loves solving problems and making things work. Right now, I am on a non-linear career path, and still waiting for the right opportunity to fully commit again. My website: https://sarusso.github.io/
I recently built a small Python library to try getting time management right [1]. Exactly because of the first part of your comment, I concluded that the only way to apply a time delta in "calendar" units is to provide the starting point. It was fun developing variable-length time spans :) I however did not address leap seconds.
You are very right that future calendar arithmetic is undefined. I guess that the only viable approach is to assume that it works based on what we know today, and to treat future changes as unpredictable events (as if earth would slow its rotation). Otherwise, we should just stop using calendar arithmetic, but in many fields this is just unfeasible...
> I guess that the only viable approach is to assume that it works based on what we know today, and to treat future changes as unpredictable events
No, the only way is to store the user's intent, and recalculate based on that intent when needed.
When the user schedules a meeting for 2PM while being in Glasgow, the meeting should stay at 2PM Glasgow time, even in a hypothetical world where Scotland achieves independence from the UK and they get different ideas whether to do daylight saving or not.
The problem is determining what the user's intent actually is; if they set a reminder for 5PM while in NY, do they want it to be 5PM NY time in whatever timezone they're currently in (because their favorite football team plays at 5PM every week), or do they want it to be at 5PM in their current timezone (because they need to take their medicine at 5PM, whatever that currently means)?
And if the number of seconds changes, such as in a batch job on a super computer, you should adjust the time of the computer first, and then adjust the billing for the job , after it completes. I asked IBM if they quantium cloud could count the time in either direction... At first they were confused, but then they got the joke.
I remember hearing at a conference about 10 years ago that Google does not make use of leap seconds. Instead, they spread them across regular seconds (they modified their NTP servers). I quickly searched online and found the original article [1].
If it works as I understood, in this setup I can see an advantage at an architectural level: in Podman containers images are stored on a per-user basis, while in this setup they would be shared between users, thus using much less disk space (if using the same base images). Besides this, I actually have the same question.
I think OP is referring to the "unprivileged user namespaces" [1] feature of Linux, which caused numerous security incidents in the past. AFAIK, this is mainly because with this feature enabled, unprivilged users can create environments/namespaces which allow them to exploit kernel bugs much more easily. Most of them revolve around broken permission checks (read: root inside container but not outside, yet feature X falsely checks for the permissions _inside_). [2] has a nice list of CVEs caused by unprivileged user namespaces.
Given that rootful docker e.g. is also prone to causing security issues, it's ultimately an attacker model / pick-your-poison situation though.
Ok, but here the OP is doing something a bit different than just rootless Docker, which is to use a "centralised" rootless Docker running as a single, non-privileged user... or am I missing something?
Your claims here are inaccurate. You can pass flags or define environment variables to get the behavior you want. Please spend some more time hitting the man pages and the guide.
> It indeed does not enforce (or even permit) robust isolation between the containers and the host, leaving large portions exposed. … More in detail, directories as the /home folder, /tmp, /proc, /sys, and /dev are all shared with the host, environment variables are exported as they are set on host, the PID namespace is not created from scratch, and the network and sockets are as well shared with the host. Moreover, Singularity maps the user outside the container as the same user inside it, meaning that every time a container is run the user UID (and name) can change inside it, making it very hard to handle permissions.
I actually went into every single line of the manuals and even discussed the matter on the official Singularity Slack.
In that blog post I wrote that it does not enforce. It is true that you can achieve some level of isolation by setting certain flags and environment variables explicitly, but this is (was?) quite hard to get working, moreover the user mapping inside the container is always host-dependant and there is just no network isolation.
To achieve something close to the behaviour "I wanted", I had to use a combination of the command line flags you mentioned (and in particular -cleanenv, -containall and -pid) together with custom-made, ad-hoc runtime sandboxing for directories which required write access (as /tmp and /home).
However, this is not the default behaviour and it is not how Singularity is used in practice by its users. But yes, I was able to achieve something close to the behaviour I wanted [1].
This said, if I am missing something, or if the project has evolved to allow for a better level of isolation by default, please let me know. That blog post is dated 2022 after all.
I agree to a certain level. However, it's hard to ensure dependencies to work in the right way without isolation. These two support tickets are a showcase of the essence of the problem: "Same container, different results" [1] and "python3 script fails in singularity container on one machine, but works in same container on another" [2]. In my experience with Singularity, there were many issues like these.
I am not sure why they had to call it a "containerization" solution. It gets a bit philosophical, but IMO containers are meant to "contain", not to just package. To me, Singularity is more a "virtual environment on steroids", and it works great in that sense. But it doesn't "contain".
The hard truth is that Singularity was designed more to address a cultural problem in the HPC space (adoption friction and push back of new, "foreign" technologies) rather than to engineer a proper solution the the dependency hell problem.
HPC clusters still use Linux users and shell access, meaning that it is up to the user to run the container: there is just no container orchestration. This means that the user has to issue a command like "singularity run" or "docker run". And since not long time ago, to let users do a "docker run" it meant to have them part of the docker group, which is a near-root access group. Just not doable.
Singularity also works more or less out of the box with MPI in order to run parallel workloads, either locally on multi-nodes. However, this has a huge price as it relies on doing an "mpi singularity run", and it requires to have the same MPI version inside and outside the container. To me, this is this is more a hacky shortcut than a reliable solution.
I believe that the definitive solution in the HPC word will be to let HPC queuing systems to run and orchestrate containers on behalf of the users (including to run MPI workloads), thus allowing to make use of any container engine or runtime, including Docker. I did some trials and it works well, almost completely solving the dependency hell problem and greatly improving scientific reproducibility. A solution like the one presented in the OP contributes in the discussion towards this goal, and I personally welcome it.
With respect to Singularity, I think they just had to name the project "singularity environments" rather than "singularity containers" and everything would have been much more clear.
