Recall Valve’s early demonstrations of Half-Life 2. The demonstrations featured quite impressive examples of enemy AI responding to player action.

When a hacker broke into Valve’s computers and stole (and published publicly) the sourcecode to Half-Life 2, it was proven that all the behaviour in the demos were in fact scripted; the AI wasn’t responding to the player at all, it was the player (demonstrator) that was acting along, making sure that he was doing what the scripted sequences expected (just like the Pleo demos).

Of course, Valve did this because of time constraints, and the early development status of the product. They eventually finished Half-Life 2, and released it to critical success. The real-life in-game AI used a combination of scripted and reactionary AI that was quite impressive in practice, and everybody forgot about the earlier “faked” demonstrations.

The parallels between the two situations are striking. So while it was dishonest for Ugobe to fake the demo, that doesn’t mean that Ugobe won’t release a good product.

As a note, from what I remember of the Valve incident, the scripted nature of the events were most obvious in one particular part. When the player runs into a house, closes the door, and barricades it, the enemy bangs on the door, decides it can’t get in, and then breaks the window to get in. In the debunking, it was shown that even if you don’t even close the door, the AI still tried to bang on the non-existent door, and then breaks the window.

Jimzip