NR2K3
AI Experiments #1
Here's a little article about one of the NR2K3 tracks and my experiments with the AI.
The Bridgehampton Experiment
Sometimes it's hard to figure where to set the AI strength when you want to race at a track and haven't run a lot of laps there. The starting Auto AI is probably too low if you've spent any time with the game and running a bunch of Quick Races doesn't always give you the results you want, probably incrementing too slowly to get you to a competitive level. It's not that much fun to lead a race start to finish and come within 5 or 6 cars of lapping the field twice. Setting the AI to 100% doesn't do it for me either--it's crazy at that level if you don't have a lot of testing and a real sweet setup. So I tried something different at Bridgehampton. I ran a bunch of accelerated Quick Races using AI only and taking the races only through Qualifying. I checked the average times against my best lap and kept lowering the AI% until it reached a level where I thought I could run. At some tracks I can wipe up the AI at 95% and at others (such as this one evidently) I'm just not ready to go that high. Remember, at this point my best lap ever was 104.408 mph and I'm comparing that speed to the AI average speed.
| AI % | High | Low | Ave | Median |
| 100 | 113.853 | 110.706 | 112.399 | 112.703 |
| 93 | 110.633 | 91.240 | 107.882 | 108.439 |
| 92 | 111.051 | 104.320 | 107.443 | 107.973 |
| 90 | 108.472 | 102.814 | 105.908 | 105.916 |
| 89 | 107.705 | 99.555 | 105.090 | 105.618 |
| 88 | 106.031 | 100.185 | 104.076 | 104.105 |
My 104.408 mph would start me in 42nd place at 100%. I thought that since 104 was within approximately 7% of the top AI speed that I should be able to go at 93%. What I found was that no AI drove that slow at 93% unless their car had serious problems. At 88% I finally reached a level where my best speed was more or less equal to the average for all drivers. I decided that's where I would put the AI for the race. It actually worked out rather well: I bested my old qualifying best by turning 105.573 but still had to start in the 14th position. And I finished 10th while really not driving that well. The AI raced pretty good but I could catch them on the longest stretch, so a little tweaking is in order to get things to where passing on the frontstretch is not a gimme. Actually I think, qualifying aside, that I should set the AI up into the 90's somewhere (if I don't mind starting at the rear of the pack).
Regarding Realism
The question begs: Could the cup cars really run this track at 110 mph today when the record qualifying Cup run was a little over 90 mph by Richard Petty back in 1964. I mean, he was a pretty decent driver you know. On a subjective basis I would say probably. Skill levels being equal, there's no comparison of the technologies then and now. Cup cars are not like Formula 1 or any of the low-slung open wheel cars but they are quite a bit different from when Petty ran. I haven't run any of the older mods to check out the difference but I expect that if Petty were racing with today's equipment he would still hold the track record and he would hit 110+ mph. That being said, it feels "real". I'm pretty much a 90-percenter across the board. That's my average speed across all tracks whether I've run one lap just to have a best time recorded or hundred of laps with varying setups. I may get close to the 100% mark on some and at others I will struggle with 80%. Five points away from the single-lap record is about average for tracks where I've really "tried" to get good.
Radt's Final Words On The Subject
(Snicker) Of course the Cup drivers could run 100+ on this track. You did it dintcha? I could do it your old Vega. Oh, you didn't tell everyone that you once owned a VEGA?