Yep, Valve also normalized microtransactions significantly through TF2.

Once again, Valve started it as something reasonable: Cosmetic options, then expanded to allow shortcutting unlocking alt weapons through $1-3 charges instead of through game progression (achievements unlocked alt weapons at first). Other companies followed suite in ever increasingly predatory ways, and Valve got worse with it too over time.

I’ll tell you something you missed:

Steam’s DRM is notoriously easy to bypass, allowing that. They also don’t force DRM on their platform, it’s entirely developer/publisher opt-in (and they are also free to add additional DRM on top if they wish), and many many releases on Steam run fine directly from the executable without the launcher running.

Edit: For the record, I pirate before I buy, buy on DRM free platforms (GOG mainly) where possible, and use a third party launcher to unify my collection across multiple storefronts and many many loose executables into one spot.

Let’s also not forget how absolutely groundbreaking Steam was for digital distribution.

I really have a hard time accepting that they “pushed” the industry rather than that they offered a platform with features that were worlds beyond what was available at the time for game developers and publishers. No one was bribed. There were no shady backroom deals. No assassinations of competitors (in fact the opposite, doing experiments with cross platform purchases with the PS3 and with GOG). There was no embrace extend extinguish, as there was nothing already existing like it to embrace or extinguish.

Also saying that they are now supporting linux and open source is ignoring a long history of their work with linux. This isn’t something new for them. What’s new is yet another large step forward in their investment, not their involvement.

Look, like you, I am concerned about their level of control over digital distribution game sales for the PC market. But from a practical standpoint I find them incredibly hard to have any large amount of negative feelings about them due to their track record, and the fact that they are not a publicly traded company so they are not beholden to the normal shareholder drive for profit at any cost. I’d love to hear more reasons to be concerned if any exist rather than “proprietary” and “too big”.

On top of that, Steam DRM is pretty notably easy to bypass, with what appears to be relatively little effort from Valve to eliminate the methods. They aren’t doing the normal rat race back and forth between crackers and the DRM devs that you would expect.

Anyway, again I’ll say: I’d love to hear more reasons to be concerned beyond “proprietary” and “too big”.

No, in fact. The nuclear lobby has been historically raw dogged by the malding fossil fuel and coal plant industries for decades. Up until recently, traditional power lobbies haven’t seen renewables as legitimate competition due to issues of scaling to meet demand, issues of location restricting where they can be built, etc.

We’ve had reactor designs ready to use the spent fuel you’re so damn concerned about for years now. Turns it into even less dangerous more spent fuel as more energy is pulled out of it (if you’ll excuse the incredibly simple summation). Incredibly efficient.

Fully researched. Risks, benefits, construction costs mapped out, maintenance costs mapped out, decomissioning costs mapped out, how long they’d be safe to run mapped out.

Every single time construction of a new plant comes up, there is a massive fucking push from the older “burn dangerous shit to pollute the air and generate power” industry to drum up fear again until the local community "not in my back yard"s hard enough to stop it.

Let me make it as explicit as possible: People like you, freaking out about hypotheticals surrounding nuclear power that they have never taken the time to understamd themselves, are a huge part of the reason why greener energy production is so slow to take off.

If green energy is so ready to take the fuck over and make nuclear obsolete, how in the absolute fuck do you explain what’s going on in Germany right now? Are they just too stupid to do things the right, safe, sustainable way that has no drawbacks at all? Or maybe, just maybe, there are still issues preventing reasonable widespread adoption of renewables, and the smog belchers want us at each other’s throats instead of at theirs?

Fucking hell. Let me know when you start accusing people of being bots or paid shills so I can just fucking block you already.

Man, we could generate some good wind power with how fast those goalposts are moving!

A good chunk of the world is still stuck where the options are coal vs nuclear for base load coverage. Of course people are going to push for the safest option for large load needs.

We’re generations away from worldwide energy needs being met entirely by green renewables and battery banks. I’ll never be against expansion of those technologies, but nuclear is an important middle step that is far less dangerous than the most widely used technologies for meeting base load (coal).

Don’t let the perfect be the enemy of the good.

What?

Coal plants are the ones that produce radioactive smog. Nuclear plants just put off steam. The radioactive material doesn’t come into contact with the clean water loop that is used to spin the turbine and generate power unless something is catastrophically wrong.

The dangerous byproduct, spent fuel rods, are stored in pools buried deep, and radioactivity is drastically abetted by the spent rods being submerged in water.

Seriously, you anti nuclear people are like anti vaxxers. It’s very minimal reading to learn how this shit works so that you can have valid critique, but no, that’s too tough.

So for those not familar with machine learning, which was the practical business use case for “AI” before LLMs took the world by storm, that is what they are describing as reinforcement learning. Both are valid terms for it.

It’s how you can make an AI that plays Mario Kart. You establish goals that grant points, stuff to avoid that loses points, and what actions it can take each “step”. Then you give it the first frame of a Mario Kart race, have it try literally every input it can put in that frame, then evaluate the change in points that results. You branch out from that collection of “frame 2s” and do the same thing again and again, checking more and more possible future states.

At some point you use certain rules to eliminate certain branches on this tree of potential future states, like discarding branches where it’s driving backwards. That way you can start opptimizing towards the options at any given time that get the most points im the end. Keep the amount of options being evaluated to an amount you can push through your hardware.

Eventually you try enough things enough times that you can pretty consistently use the data you gathered to make the best choice on any given frame.

The jank comes from how the points are configured. Like AI for a delivery robot could prioritize jumping off balconies if it prioritizes speed over self preservation.

Some of these pitfalls are easy to create rules around for training. Others are far more subtle and difficult to work around.

Some people in the video game TAS community (custom building a frame by frame list of the inputs needed to beat a game as fast as possible, human limits be damned) are already using this in limited capacities to automate testing approaches to particularly challenging sections of gameplay.

So it ends up coming down to complexity. Making an AI to play Pacman is relatively simple. There are only 4 options every step, the direction the joystick is held. So you have 4ⁿ states to keep track of, where n is the number of steps forward you want to look.

Trying to do that with language, and arguing that you can get reliable results with any kind of consistency, is blowing smoke. They can’t even clearly state what outcomes they are optimizing for with their “reward” function. God only knows what edge cases they’ve overlooked.

My complete out of my ass guess is that they did some analysis on response to previous gpt output, tried to distinguish between positive and negative responses (or at least distinguish against responses indicating that it was incorrect). They then used that as some sort of positive/negative points heuristic.

People have been speculating for a while that you could do that, crank up the “randomness”, have it generate multiple responses behind the scenes and then pit those “pre-responses” against each other and use that criteria to choose the best option of the “pre-responses”. They could even A/B test the responses over multiple users, and use the user responses as further “positive/negative points” reinforcement to feed back into it in a giant loop.

Again, completely pulled from my ass. Take with a boulder of salt.

I would hope, but I’ve been burned enough to not assume big companies are doing things the sensible way.

Sorry buddy, but that capitalistic framework is where we all have to exist for the forseeable future.

Giving corporations more power is not going to help us end that.