Upgrade your account to a Community Supporter account and remove most of the site ads.

Reply to thread

Message: <blockquote data-quote="EzekielRaiden" data-source="post: 8922545" data-attributes="member: 6790260">That wasn't my intent. I genuinely couldn't make sense of the seeming contradiction.Because I'm not doing that. I genuinely don't understand how you get only defining interesting results and actually swingy results. For there to be unrestricted "swing," as I understand the term, there must be an extremely wide latitude--including plenty of results I (and many others) consider to be terrifically uninteresting.As I said in my other post, randomness is almost always unsatisfying in the long run. It's much like (for example) AI efforts at mimicking classical music. They can almost always create interesting, even beautiful passages, sections that do something clever or unusual...and these lumps float in a soup of random garbage. The more you emphasize the randomness and "swing," the more unsatisfying and unintelligible the product becomes (the same applies to basically all "generative" AIs; Google Deep Dream shows what happens when you uncap the randomness and let the AI see whatever it "wants" to see, for instance.) You always have to apply restrictions to the probability to create something which gives the kind of satisfactory results most players seek today: either by cleaving out possibilities that could happen but which wouldn't be interesting, or by restricting the "swing" so that uninteresting things couldn't actually happen. The two ultimately cash out exactly the same way. Analogically, one says, "Sure, the dice said X should happen, but I don't like X, so X didn't actually happen." The other says, "Since I don't want X to happen, X isn't even something the dice are allowed to produce." Both things eliminate X as an option. The only differences between the two have nothing to do with whether they permit unrestricted "swing" (because neither does), but rather whether they give the impression thereof. As I have been quite clear in many places, I oppose the former because of the false impressions it gives, but that's irrelevant to the question of whether they permit unrestricted "swing" or not.Hence: How do you preserve unrestricted "swing" while narrowing things only to "interesting results"? Randomness unconstrained produces white noise.</blockquote>

Verification