My belief that AI may have helped Blizzard plan, set up and smoothly launch WoW Forever is inferred and based on the poor quality of their other recent launches. WoW Classic released in 2019 and was a laggy mess for the first few days. Blizzard recently said WoW Forever is their biggest launch in history for peak concurrent players, bigger than Classic or any expansion they've released including Burning Crusade and WotLK.

Blizzard has long been derided for its poor performance when handling massive player events, showing worse performance in that regard than large private servers such as Nostalrius.

The biggest technological advancement in the past couple years that could help explain their sudden and dramatic improvement is AI, in my opinion. I think it's fantastic if AI is indeed a large part of the reason for their rapid improvement.

This is the type of change I am referring to when inferring that AI may have helped develop the changes that enabled such a massive improvement over previous launch events.

I am not a Blizzard employee and don't have insider information. I'm just impressed with their improvements and hopeful it was at least partially enabled by their use of AI to help ideate, design and implement.

There's a lot of new players that never played wow in any iteration, so I would think so.

> I actually love Q4 as it packs so much into the game. I guess it’s a love or hate game.

I guess it is, and if we're being honest I'm a certified hater.

I played all of the Quake games online, a lot, at their corresponding time of initial release. First on software, then accelerated on 3dfx Voodoo cards, then later on NVIDIA RIVA TNT and the original GeForce cards (to tie this into some of today's other big news).

I played untold hours of multiplayer online netQuake, Quake II and Quake III Arena and have my own semi-controversial opinion that Quake II was actually the best of them all (most people tend to either prefer the original, or III). But you may be the first person I've ever encountered online or IRL who has professed love for Quake 4. (I'm sure there are others out there, I just haven't run into them).

I barely played any of them in single player, to be fair, so that could very well be a factor in the difference of opinion(?).

A few things:

- Splitting your community is a concern. The classic example is multiplayer DLC: if not everyone can access it then it will risk low adoption.

- Presales and first week sales are hugely important to financing projects. Not many games achieve a long tail.

- Developers are expensive. Paying a team of artists and designers and programmers demands that they generate enough revenue to make it worthwhile.

AI opens this door because:

- Its capacity to be constrained fits neatly into mod frameworks.

- It's incredibly cheap to operate, compared to the human team.

- Users are comfortable with its SaaS model and so won't be deterred by a subscription model.

Most modding communities see it as a creative hobby, something that they do for the love of the games and communities. Paid mods are generally not something they're very receptive to. These communities are mostly people who are into retro or indie games and reject modern gaming's various monetization methods.

There are exceptions of course.

"and at no cost"

Leaked source code,flame wars, toxic fanbase making shitstorm why work with this modder but not that one, same idea as secret internal plans .. and unclear legal situation in general with hiring people om the other side of the world is what come to my mind why this is not happening more often. But it does happen sometimes, usually the modders get a individual offer from the company.

Less like a carpenter using a hammer and more like a Sorcerer's Apprentice using brooms. Your give them instructions, and they mostly take it from there. Hopefully your understanding of those instructions aligns with theirs, and hopefully when it starts to diverge you'll know how to course correct.

AI is fundamentally different to other tools because it directs itself for the most part (i.e. has agency) - the user typically only provides direction, that too usually vague direction. It's like calling a vending machine a tool for producing soda.

No, if you used an LLM to do something you deserve zero credit. You applied no skill, you didn't do anything. Anyone could do the same.

To be honest, you make a good point. The problem is generally that AI erases many of the tells of a quality work, because even bad works tend to carry the exact same level of language. When I see that the entire description was AI-generated, I wonder how many of these points were actually inspected by a real person or just hallucinated by the model. Surely there is some quality to it, but I feel like I would need more qualifiers than "finished" to know if it's worth trying, you know?

It's not exactly analogous to a hammer, even though I agree that it is a tool. If a carpenter got the framing all wrong, the hammer wouldn't fix it. If a prompter got the instructions all wrong, or didn't supervise enough, or didn't verify enough, the AI still makes something that appears to pass initial muster.

are... you arguing for, or against his comment with yours?