wasfgwp

Do you know how many slaves were black in the Roman empire

In the Western Mediterranean? Most likely diminishingly few. It’s hard to imagine why would that might not had been the case.

Equating North Africans who genetically, culturally more interlinked with other Mediterranean people with Sub-Saharan Africans is not exactly right. After all before the Indo-European invasion Europe was almost exclusively settled by middle eastern immigrants (and largely remained so in Southern Europe genetically).

It's not a stretch to imagine

It’s not a stretch to imagine a lot of things but there is no evidence that was true to any significant degree (besides the demand for slaves in Roman Egypt was likely substantial enough to absorb most of the demand)

Italy was Roman though. Or rather by making Italy Roman (to such an extent that other Italians would be willing to fight a war just for the right to be treated as other Romans) was basically what enabled their success by giving them a huge pool of manpower (and most importantly of people who had a vested interest in the success of the Roman state). Other ancient states like Athens were inherently exclusionary which put a hard limit on their power projection capacity since they depended on fickle allied or subjugated states to fight their wars.

I wouldn’t call it an evolution. The game might be good in its own right but its different in some very fundamental ways besides the historical setting. Especially on the competitive level.

At the very top probably not, at least these days. But historically low APM players were often able to overcome that through better decision making and get to the top 4 or so. That’s generally still the case outside of the top 4-8.

Now the meta is quite established and it’s much easier to get better at decision making through practice than becoming physically faster.

It’s very complex though while the factions are mostly symmetrical there are over 30 of them and they all have marginal advantages/disadvantages. Games are more usually than not won through optimally utilizing them. So in tournaments drafting became very important since using the same civilization twice or by opposing players is almost universally disallowed. So to be successful you need to outwit the opponent by picking the right faction(s), be capable of making the optimal strategic decisions and be very fast/good at micro (which is also probably more important than in most other games since arrows are not heatseeking missiles and pathfinding is extra wonky)

Well it is a bit surprising that 3.1 flash-lite could be better than deepseek-v4-pro (cheaper output and way cheaper cache so might cost less for quite a few use cases).

They are not anywhere close according to pretty much every benchmark (even v4-flash is considerably ahead and its way cheaper than flash-lite). Maybe tuning prompts/tools/etc. might be useful?

And most potential buyers would likely prioritize buying a PC before they consider getting a Steam Deck so it’s likely even more sensitive to price increases.

Well in relative terms that hardly matters for consoles due to obvious reasons.

Developers of AAA games for PC can only (or are willing to) do so much to make it run acceptably on what’s a very underpowered PC compared to their target market.

That’s a somewhat privileged view though? Most people are just unwilling to pay >$800 for a handheld console (or can’t actually afford it).

Presumably most potential buyers have PCs already that they use for gaming so it’s a luxury to spend that much on a Steam Deck.

Hyperscalers would generally be expected to sell compute to anyone who is willing to pay assuming that supply/demand for datacenter hardware evens out at some point that should become true again.

Isn’t OpenCode just reselling Deepseek’s tokens over the same API? For v4-pro the prices from other providers on OpenRouter are way higher so what else could it be?

Opencode Go already seems extremely unprofitable if people use up at least 30% of their monthly tokens so its hard to imagine they’d go with several times more expensive inference provider yet show Deepseek pricing on their page.

I think that pretty much all Chinese providers do that?

And Codex obviously allows that as well. Anthropic is pretty the main exception.

Presumably people want to hear the opinions of other people and still believe that they might learn more from the their experience than by exclusively interacting with AI models?

The alternative of that first impulse should be what exactly? Telling AI to research Medium AI drivel articles (since nobody seems to be posting long/medium form content anywhere else these days) to figure out what’s the best approach to building an AI harness?

BTW just my personal observation but I generally find AI notoriously bad at building any AI applications compared to standard software (which is not surprising given the scarcity of high quality training data)

It was certainly not flat. There was a lot of volatility.

Not sure where you've read that it was

Well.. I looked at the charts. The bottom was few months after Pearl Harbor and pretty much straight up after that. It certainly did much better after US joined than while it was staying out..

Not really. Or rather not on the same level. Both the government and the military were reasonably competent when it came to winning the conventional wars back then. Both failed at handling the insurgency phase (the political leadership to a larger extent).

Now? Who knows, I guess the military is doing reasonably fine given the constraints they operate under but I doubt competent is the right word.

What is clear is that Bush (in combination with his administration) was pretty much Napoleon compared to Trump, Hegseth & co.

They occupied the country and mostly/almost did accomplish that. 2002 to ~2005 were reasonably clam.

Then their failures to bulld a stable government brought the Taliban back.

Also USA is very sensitive about the other side's civilian deaths.

Well depends what you compare against. Also its generally not worth wasting precision munitions (you have a limited amount of them to start with) on non military targets.

The stock market did great during WW2, though.

Of course this time its not dominated by industrial conglomerates which can retool their factories to manufacture military equipment.

Bush won the conventional war in Iraq and generally the US military was quite good at that part (same in Afghanistan or the first Gulf War). They had no clue what to do after the war was won and made a complete mess of it.

Trump hasn’t got the slightest idea how to even get as far as Bush did..

The majority (60-70%) of Europeans still commute by car. Distances might be shorter and their cars might use less fuel but on the whole its still more appealing option for most people (despite quite a few countries trying to actively change it by making it more expensive and leas convenient).

https://ec.europa.eu/eurostat/statistics-explained/index.php...

To be fair if you have some background in data science (I think the main thing is being able to verify its not showing nonsense) it shouldn’t be too hard to build a bespoke solution.

do it where it is possible.

So lets only boycott small and inconsequential companies like Mullvad that are easily replaceable. However not companies like Apple, Intel or Nvidia etc. whose CEOs have expressed their personal admiration of Trump and supported him financially because it would not be very convenient?

To be fair that seems to be reasonably rational i.e. anthropomorphising lawnmowers is a fool’s errand while its feasible to actually make a difference in cases like this

Their CEOs and the tech megacorps have been openly supporting Trump and financing both him personally and his political projects. There is no ambiguity in that at all.

I don’t remember Tim Apple or any other tech CEOs almost literally groveling at the feet of any other president besides Trump.

They don't have to also be the main supporter of some ethnic cleansing party

True OTH that party is mostly inconsequential and has no direct power. Tech companies enabling and legitimizing MAGA by openly supporting the movement arguably results in way more damage on aggregate even if they are only financing through outright bribery.

To be fair arguably there is no other country in the world which benefited as much economically from its interactions with Nazi Germany and face so little consequences for it.

By any chance are you using any Apple products and are you a supporter of Donald Trump? Because if the answer is yes and no singling out Mullvad seems be rather hypocritical?

Based on his actions Tim Apple openly is an ardent supporter of Trump. So effectively everyone here who has a MacBook was supporting a company run by by a pro MAGA ceo...

Of course greed and corruption might be considered a mitigating factor by some people (instead of seemingly doing it purely due to one's personal convictions and without any coercion which is presumably the case here

You are assuming the interest of people making those designs were aligned with the interests of consumers. They obviously were only to a very small degree since there were almost no incentives for Soviet companies to produce anything that wasn’t complete crap. Consumers had no choice since even the crap products they produced were hardly ever available to normal people anyway.

USSR just wasn't rich enough

To an extent by choice. They really didn’t utilize the resources they had optimally.

Usually to maximize its funding the company would move its HQ to the US and if they are lucky have an IPO there and eventually become effectively American after a few years (e.g. Unity)

beats claudes sonnet

Based on benchmarks which don’t mean that much these days.

models is gone as of the last few weeks.

Yes, that’s exactly what people were saying after every major release for the past year or so. It’s always a couple of weeks away

high as the second best general purpose model

According to benchmarks which are gamed to the extreme these days. Trusting them blindly isn’t exactly rational either. They don’t necessarily translate that well to real world tasks

It’s obviously not “distilling” as such but there are reasons why Chinnese models are consistently several months behind OpenAI/Antropic