LLMs are actually useful, people are willing to pay for access to them, and they do genuinely enable things that were unrealistic or impossible before. The advances in image, video, and sound models since 2020 are also striking but likely won't be as transformative as LLMs.
That being said, I don't think it's unlikely that we'll see a plateauing of progress followed by a strong crash/correction in the market a-la dot com. The Allbirds situation absolutely has echoes of pets.com.
I also feel that commodification is coming for models, training/inference hardware, and software (e.g. CUDA), as it has for nearly everything else useful in tech. So I expect valuations driven by unique advantages here to be eroded over time (Think Sun and SCO after Linux on cheap x86 servers became the norm).
Granted I skimmed the video and checked the slides, but as far as I can tell the presentation doesn't mention Perforce once?
If you're marketing a new version control system for game development, then you really need to be talking about how it compares to Perforce and not Git.
The fact that Git is sub-optimal for many game projects is well established and is a key reason why Perforce is the dominant vcs in the space.
> since modern Germany is a very synthetic unification of what were historically very different sovereign nations
This is fairly revisionist. Germany is no more synthetic than most other modern European countries in that regard. The lands of modern Germany were more culturally and linguistically unified than say, France or Spain (which had more diverse minority languages that were suppressed, e.g. Catalan, Basque, Occitan, Breton).
Virtually everything that is now Germany (and lots that isn't) was part of the same polity, the Holy Roman Empire, for roughly 1000 years before being dissolved by Napoleon in 1806. Yes, it was highly decentralized even for the time, but the fact that there was a common language in the core lands was not lost on people at the time, and a gradual informal standardization of the written language was already taking place during this time.
So Germany's unification in 1871 was essentially re-unifying lands that had been part of the same state for centuries before, only this time they did it under the justification of the then-fashionable concept of the nation state and with a centralized power in control of the whole thing.
Unfortunately for everyone, that centralized power was Prussia.
I'm going to assume that you're not being intentionally obtuse, so I'll clarify what people are saying.
Germany didn't and still does not have meaningful immigration enforcement. For fairly understandable reasons, German asylum law is quite permissive. You get benefits from the government (read, from the German taxpayer) as soon as you submit an asylum claim even if it ultimately gets denied. Processing an asylum claim takes a long time, and there are many ways to delay it. It is very hard to get deported by force from Germany.
The net effect of this was that a huge number of young men who were essentially economic migrants came to Germany in the last 15 years on dubious asylum claims by misrepresenting their age, their national origin, the situations in their home country, and so on. A lot of these people are perpetually unemployed, not only due to the complicated nature of getting work permission as an asylum seeker, but also because they lack the language and professional skills to work the in-demand jobs in Germany, and because, well they continue to get government benefits.
This is on top of the fact that the German social support system is already strained from age demographics, and that large groups of young men who come from misogynistic/chauvinistic societies and are perpetually unemployed behave rather predictably.
In addition, by allowing this abuse of the system, the skilled legal migrants also suffer as they are heavily taxed to subsidize non-working "asylum seekers" and growing anti-migration/anti-foreign sentiment gets directed to them as well.
So in conclusion, refugees and migration are not "totally different things". In fact, in the case of Germany they are two deeply intwined topics that cannot just be dealt with in isolation.
If you think US foreign policy decisions are, or ever have been, made primarily on the basis of preventing the suffering of civilians in other countries then I have an Aerospace & Defense Sector ETF to sell you.
Snark aside, the real answer is that in the past it was infeasible and undesirable due to the geopolitical situation with China and the USSR. Any major operation there would have been a Cuban missile crisis level provocation. After North Korea successfully built nuclear weapons, any sort of regime change games like the US has done in other countries became totally out of the question. There may have been a brief window in the 1990s where it would have been possible, but it still would likely have resulted in mass civilian casualties (look at where Seoul is on a map).
The only "good" realistic scenario for North Korea in the next 20 years is that they gradually reform their system akin to what China did in the late 70s, early 80s.
It would be cool technology today if we lived in a system that prioritized human dignity and general welfare over the desire of a handful of people getting obscenely wealthy.
Highly dynamic languages existed for decades prior to 1995, Python was not particularly innovative in its features at the time. There were also countless languages more feature-rich than C being used for development at the time.
The biggest change that happened was that hardware kept getting better and it became feasible to use garbage-collected languages everywhere including really inefficient implementations like CPython.
That being said, 30 years later Python is still slow as shit even compared to other dynamic languages and runs into all kinds of scaling issues when used for anything serious. And everywhere that performance matters, software continues to be written in typed, compiled languages including C (but also C++, Rust, Go, etc.). Even in ML, Python chiefly acts as a thin wrapper and glue language for high performance CUDA libraries (aka C and C++).
So your historical analogy is mostly anachronistic.
> it mostly boils down to recklessness of developers
I disagree. I think in big tech and the corporate world, it boils down to the organization fundamentally not valuing security and punishing developers if they "move slow", which is often the outcome when you maintain a highly security-oriented process while developing software and infrastructure.
When big leaks happen, the worst that occurs is that some trivial financial penalty is applied to the company so the incentive to ignore security problems until you're forced to acknowledge them is high.
Modern Germany will collapse in the face of any war that would justify conscription, so it's all a moot point anyway.
The country is already in a slow burning crisis due to the political and economic results of its demography, and a war coming to its own soil will send the walls tumbling down.
The philosophical justifications sound nice and all.
The thing is that when you have a huge non-citizen percentage of the population that is actively drawing taxpayer money out of the state to the point where the social welfare system is beginning to break down, and you have the working citizens of the country being taxed at 50% or more to support that during an escalating global cost of living crisis, you have effectively destroyed the social contract around citizenship that permits this system to function. For the massive aged population that's drawing retiree benefits, there's at least the justification that they paid into the system during their lifetime, even if the equation that makes the retirement system work increasingly doesn't work anymore.
Now the young people are being told to go die to keep that system alive. I wouldn't be surprised if most don't.
It would be equality if there were a law forcing women to have children during a war. Which is insane and no one would support it.
But young men maybe dying after being forced to fight against their will? Completely fine.
It's honestly just very telling how in modern Western egalitarianism, gender essentialism is factually wrong and evil unless we're explaining why men need to die for their country.
Virtually every population outside of Sub-Saharan Africa has Neanderthal and Denisovan DNA between 1-4%. This includes all of Eurasia, all pre-Columbian American populations, Aboriginal Australians, Papuans, etc.
I pretty much declared streaming show bankruptcy after sitting through Severance season 2 last year.
I know a lot of people liked it and maybe I'm just cynical, but to me it seems like every "serious" streaming show eventually falls victim to the "stretch a 2 hour movie's plot across a 12 - 16 hour season" strategy. They know it works because enough people binge watch or feel compelled to finish a series they've started.
At this point, if I'm watching a show then it's something where the episodes are sufficiently satisfying self-contained stories (e.g. something like Star Trek, X-Files, sitcoms). If I want something with a more involved plot, then I'll watch a movie. These formats are better because the limited runtime requires the creators to be intentional about what they dedicate screen time to. Meanwhile in a modern "story-driven" streamslop show it's painfully obvious when they're just padding out the runtime with fluff to make it to 8 episodes.
Of course there are exceptions to this, and there are stories for which a miniseries or a long-form series is the ideal video medium to convey them. But what happens so often is that you get 1-2 seasons of compelling storytelling followed by N more mediocre seasons that keep getting made because enough people keep watching. And the latter are just not worth the time investment.
LLMs are actually useful, people are willing to pay for access to them, and they do genuinely enable things that were unrealistic or impossible before. The advances in image, video, and sound models since 2020 are also striking but likely won't be as transformative as LLMs.
That being said, I don't think it's unlikely that we'll see a plateauing of progress followed by a strong crash/correction in the market a-la dot com. The Allbirds situation absolutely has echoes of pets.com.
I also feel that commodification is coming for models, training/inference hardware, and software (e.g. CUDA), as it has for nearly everything else useful in tech. So I expect valuations driven by unique advantages here to be eroded over time (Think Sun and SCO after Linux on cheap x86 servers became the norm).