marginalia_nu

There is shovelware at any level, but at the same time I think indie games have been a more consistent stream of good memorable entries than big budget studios the last 10 years.

I really have to rack my mind to find think of anything that I feel has been memorable the last decade, and even then half of them are AA titles (and many Japanese, dunno if it's the gambare ethos or what's up with that, maybe I'm a closet weeb). Nioh 2. KCD2. Expedition 33. Yakuza: Like A Dragon but not the sequel. The Elden Ring DLC. Prey 2017. Doom 2016 and Eternal. I guess Stalker 2 has been getting pretty good too.

For good indie games I can drop dozens of names without stopping to think. There's Hollow Knight and Silksong, Hedon, The Drifter, Lunacid, Cruelty Squad, Voin, Paradise Killer, Blue Prince, Fallen Aces, Not For Broadcast, Heaven's Vault, Chants of Sennaar, Dead Cells, Blasphemous, Immortality, Axiom Verge, The Blind Prophet, Cross Code, Cultist Simulator, Astalon, Dread Delusion, The Witness, The Signifier, ... I can keep going.

(I authored that blog post)

I think it kinda depends on the stage of the development. If you have an application already, flame graph whack-a-mole is the normal starting point. This is the correct way to go about it because of Amdahl's law. Random optimization is extremely unlikely to pay off in a way that leads to measurable application wide performance improvements; no matter how big improvements you find for a function that's 0.1% of the run time, you'll never see more than a 0.1% improvement of the run time.

Bottoms up optimization is what you reach for if you have a known hot spot (from a profiler flame graph), or you're deliberating design choices when building something from scratch (where you can be reasonably sure performance will matter).

Of course all of this is exponentially harder for database-adjacent work since you have os and hardware level caches, possibly a buffer pool of your own, and in production you usually operate in fickle semi-resident states that are extremely hard to replicate in a benchmark scenario. So you end up with a lot of benchmarks (cold cache, hot cache, cgroup memory constraints) hoping that sort of fixes it.

I think overall there's been a huge amount of regression in discovery for music but also like... most other things. Recommendation algorithms tend to show you more of what you already like, which makes it incredibly hard to discover new things you might also like. Also creates a huge popularity bias where what's popular gets more popular and what isn't doesn't stand a snowball's chance in hell.

One of the lost joys of the past is reading a print magazine that has no idea who you are and discovering articles not written for your demographic, but that regardless expose you to new perspectives and ideas, maybe you find a new hobby or band or whatever, and enrich your life forever. That stuff just isn't happening anymore.

Picard is both the writing and Patrick Stewart. Picard would not be the same character if they'd cast Kevin Sorbo to play him[1]. Picard without Stewart's presence and gravitas would be something else entirely.

[1] Sorbo's portrayal of Dylan Hunt is basically just taking his Hercules role and putting him on a space ship and I have no doubt it would look the same on the bridge of the enterprise.

Proton is a for-profit owned by a non-profit foundation, and Yen is the majority shareholder of the company and board member of the Proton foundation. They were ostensibly also very financially stable. He had every opportunity to be a Picard. He chose to be a Ferengi instead.

Proton had up until two days ago ostensibly cornered a niche of the market where they could basically collect taxes off security conscious nerds and other people wanting out of big tech. This wasn't a start-up with 3 months runway. Proton was by all accounts very financially stable.

The literally only way they could imaginably fuck up what they had was if they squandered the goodwill and trust of their users.

[flagged]

They had a safe stable business with a loyal user base in a market niche they completely owned, could have basically just sat on their asses and de facto collected taxes from paranoid nerds and journalists for decades.

I just don't understand why they'd want to trade all that away for a chasing the same nebulous "something something AI" business plan as literally every other company in the tech industry right now. The odds they'll be competitive in such an endeavor and see a fraction of the success they had again is minuscule.

[citation needed]

Even still, there are gradations of using AI. Yen seems to have been on a crusade to get everyone to vibe code the entire thing. Not a lot of high performing software engineers signing off on such practices either.

You treat your team well because your highest performers are the ones with most options and the ones that will leave first when you don't. If productivity is bad before you pull something like this, you won't like what happens when all you have left is a team full of demoralized underachievers.

According to the manifesto, it seems to be that hopping aboard the AI train has inexplicably not materialized exponential revenue growth, and since it can impossibly be that AI has been oversold, the only conclusion must be that employees are dragging their feet and must be forced to go full send with vibecoding for the destined profits to arrive.

I've literally had videos like this on my feed almost all year. Idk, maybe it is about to pop, but the amount of crying wolf about it makes it hard to take it seriously.

Yeah chasing velocity like that is for startups. Proton already has a product, a competent team, and (had) a lot of soft capital in the form of a good trusted name and goodwill.

That's enough momentum to coast on for a very long time, especially since their whole deal is that they're owned by a nonprofit and don't even have shareholders to pressure them.

They really had to nothing and they'd won, the only way to lose in a situation like that is... exactly this. Andy could have probably just sat on his ass for 20 years like Gabe Newell and probably been able to buy a similar yacht at the end of it.

Can't imagine the company surviving if Andy is allowed to stay on as CEO. A company like Proton lives and dies by its reputation, trust, and goodwill. Given their mission and brand, this is spectacularly tone deaf and jeopardizes everything they've spent over a decade building up.

Only thing that might save their face is kicking Andy and following up with Japanese levels of corporate apologizing.

If anything the expectation of incredible improvements just over the horizon makes an argument for not adopting AI or at least doing so very judiciously.

Why would you spend dollars on tokens today if you expect the same tokens will get you something vastly superior in a few months, or that they'll do the same job for nickels on the dollar? Do you just hate having money or what is the problem?

It's great when you have it set up but there's so a fairly gnarly amount of ritual magic needed to set it up. I kinda wish it was first part integration, especially given the extremely lackluster 3D level editing abilities of Godot itself.

Godot feels very much held back by how much stuff you need to do in 3rd party apps with flakey or nonexistent integration. Software developers often get shit for the learning curve of IDEs, but nothing has the "damn girl, you live like this?" energy as open source gamedev tooling. Even something pretty basic like doing texture painting in Blender? Jesus H. Christ. Makes The GIMP seem more polished than anything Apple put out at the peak of its polished days.

It all reminds me of my favorite Asimov quote:

> [...] when people thought the Earth was flat, they were wrong. When people thought the Earth was spherical, they were wrong. But if you think that thinking the Earth is spherical is just as wrong as thinking the Earth is flat, then your view is wronger than both of them put together.

Source: An image on twitter.

Mostly economies of scale. They have enough data to do the statistics to identify bad actors, and enough bandwidth and servers to facetank the request volume.

I don't understand why this can't be on a single page. Makes it really hard to refer back to earlier "chapters" as you get new information when it's a slideshow. Can't search either.

Seems like we're needlessly reintroducing limitations from the paper medium that don't exist in digital media for weird aesthetic reasons?

Won't address start up time since it's so affected by which libraries you pull. Slow starts are usually related to dependency injection frameworks like Spring or Guice, which admittedly is what most nontrivial projects will tend to use.

Performance depends a lot on how you write the code. Inefficient Java code is often quite slow, but you can get surprisingly close to C++ speeds with Java in many cases, but that involves a lot of byzantine rules of how to write code so that it e.g. omits bounds checks. More realistically if you don't pay any attention you're probably running at like a 25-50% speed penalty.

Java has a leg up on statically compiled languages in that it does automatic PGO based on real run-time data which helps bridge the gap to a surprising degree. GC is also a lot more resistant to memory fragmentation from repeated small allocations than malloc, which can help as well, not just from its generational approach but also e.g. TLAB space.

It's a bit held back by not having decent access to SIMD intrinsics. Maybe will resolve itself when the Vector API gets finished baking eventually.

Yeah this is why it's such a pain in the ass to produce dairy products. Have to find a wild pregnant cow and capture it, then throw away the calf before it drinks all the milk.

Just comparing hours hides the reality of what the time outside those hours looks like. When 50h weeks were the norm, you generally had a stay-at-home wife making that whole equation possible.

For some reason when we started sending women out to work, the work they had been doing at home stopped being counted as work. But of course it is work, it still needs to be done. As a result, the modern 50h weeks can be anything between 60-80h weeks depending on your comfort with living in squalor, and the presence of children, it's just we for some reason only count the part you get paid for.

There are many cases mammal species that almost never breed in captivity. Giant Pandas and Elephants are perhaps the more well known ones.

A reminder that technically correct is the best kind of correct.

The Great Depression began in 1929.

To be fair, depending on when these things were written, there was a lot of shit going on in Greece and Rome that merited such a take, including invasions, plagues, civil war. And the '20s[1], '30s[1], and '40s[2], and '50s[3].

[1] https://en.wikipedia.org/wiki/Great_Depression

[2] https://en.wikipedia.org/wiki/World_War_II

[3] https://en.wikipedia.org/wiki/Cold_War_(1953%E2%80%931962)

I don't think there is an obvious contradiction between what you're saying and what the post you're responding to is saying.

Skeumorphism isn't necessary for consistency in user interfaces, but it generally ensures a level of consistency that is honestly hard in modern interface design.