Is this supposed to be depressing?

Man, after reading that, I feel like I was just watching a VH1 riches-to-rags-to-riches story.

Man, right at the start of the article:

Today is March 21, 2003--the start of the second war in Iraq, which is a fitting parallel to the battle Newell is about to start with the announcement of Half-Life 2 and its release date.

What the hell? No it's not! Have some sense of proportion!

This article underscored a lot of things that I learned at NITI about development:

- The worst thing to do is have a huge chunk of work to do with nothing to show for it. Get something you can release and then build on it. Obviously for a game that people will buy and play once, that won't work at all, so stay away from games if you value your sanity. (Counterexample: NWN, whose first edition was pretty mediocre but was obviously just a first step.)

- Don't try to build all the tools yourself. That's time you could be using to make stuff to sell.

- For God's sake, pad your estimates to customers. But not to employees - give them the real dates. The schedule is the last place to play head games.

- Organize yourself around lots of little releases, so that when features slip you can just push them back to the next release. Again, this wouldn't work for a project like Half-Life, so the key here is - avoid projects like Half-Life if you value your sanity.

The EA Sports games are a model that could work like this. You get one a year, so you add small features to the game and integrate them often. Every month you have an essentially playable game that's a little bit better than last month's, and when it gets close to the release date you just freeze this month's and call that NHL 2k5. (Hmm, aren't you working on the EA Sports games?)

Valve's design *could* have worked like this too, if they were willing to focus on selling the tools for people to create the Counter-strike and Opposing Forces games: get your engine working and release it to let the fan-made games make mistakes in interactivity. Release a slightly better engine every month as you take your time in building up your own story and then eventually release it as the first official Valve game with the already-released engine. You lose the hype and excitement of unveiling your new game to the public in one shocking blow, but you get a more stable and less stressful process over the long term. It's less sexy, but it works better.

That's essentially what Bioware did with NWN, BTW - the first version came with a pretty mediocre campaign, and then they divided their time between improving the engine and making more impressive games with it. I guess MMORPGs could fit this model, too, as would pure strategy and action games. Really, the only thing that kills the "release early, release often" plan is storyline.

indeed

I'm working on NBA Live 2006. I agree, the small feature addition model of development could work well for a sports franchise title like this. It's not the most efficient in terms of pure productivity, but it's totally worth it for the ability to be flexible in managing the feature set and not getting stuck with massive crunch times at the end.

Incremental design wouldn't work as easily for a game like Half-Life, but it wouldn't be impossible. It would be a matter of planning out your design goals in order of storyline importance, and being willing to cut some levels in the middle that don't have strong additions to the storyline. Or alternately, tell your writer(s) to make a great story with the full set of locales, but also have backup plans for making a continuous story if one level is dropped. Could we change some dialog and scripting in the levels around the cut and still maintain a consistent, solid storyline? Know ahead of times which levels are key for plot, which are key for gameplay, and plan for compromise.

Of course, the real issue in the HL2 story seems to me that they simply didn't have someone designated as a whistle-blower. It makes zero sense to me to have a producer with creative control making decisions about scheduling. The two goals are completely at odds with each other. You need a manager who's job is completely separated from deciding if the game is good enough, who's not going to get sucked into a giddy-schoolboy mentality of, "This game is going to be so great, we just need to push harder to get these ten more features in!"

I mean, hearing how late they finally had the guts to tell him that they weren't going to hit that schedule is ridiculous. I wish that managers were perfect in their time estimates, but if you're unwilling to have the guts to tell producers that they need to choose between a ship date and feature cuts, then you're just killing yourself (and quite possibly killing the game).