Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Bumber]

Walking simulator after 70 generations:
0   sofuho fuaame   3.64
1   sofuho fuaame   105.89
3   oonede eiaenu   107.01
5   oonede eiaenu   107.07
6   oonede eiaenu   107.53
7   soneme eiaeou   108.41
8   nonede eieequ   109.19
13   ninede eimequ   110.17
20   qinedo eayequ   110.20
35   soticu eeaipi   111.97
54   sipeee fazapi   112.91
64   pipude hazali   212.96

Interestingly they learn to take one step fairly quickly. Then they spend a long time not accomplishing very much.

[Arx]

I will not forsake the bay12 world!


Top Scores:
#1: 198.75 d:195.45 h:-8.06/9.36m (gen 1609)
#2: 194.42 d:190.65 h:-8.25/8.43m (gen 1743)
#3: 194.37 d:190.97 h:-7.85/9.28m (gen 1318)
#4: 194.27 d:191.24 h:-7.98/8.61m (gen 1642)
#5: 194.19 d:191.03 h:-7.89/7.98m (gen 770)
#6: 194.01 d:190.23 h:-7.79/9.24m (gen 1645)
#7: 193.96 d:190.43 h:-7.85/9.42m (gen 1527)
#8: 193.93 d:190.92 h:-8.19/8.32m (gen 1638)
#9: 193.91 d:190.96 h:-7.88/9.35m (gen 1499)
#10: 193.86 d:190.88 h:-8.19/8.73m (gen 1620)

Edit:

Well, it's been over 600 generations and nothing's been able to pop the cap. Gonna call it.

[sambojin]

Bloody hell. Is that on Earth gravity, fixed track, bay12 world?

You Sir, win an internet.

[Aklyon]

Post to cars.

[miauw62]

Bump I guess.
This game gets pretty easy.



mutable terrain, too.
the big increase in averages is when i changed the mutation rate to 10% and mutation size to 40%. as you can see, cars that go too far just end up falling off when the map stops at, i assume, 250m. after that it's just a question of speed. the cars i have right now really just depend on the terrain allowing them to fall off (e.g. no vertical walls anywhere)

[scrdest]

Well, that's frustrating. I had a guy who beat the 120 Pothole and then survived the cliff by bouncing back at Gen 82 but they almost immediately regressed into terrible strategies.

As I was typing this a guy got to 148 but it seems like it's Amateur Hour again.

[Reelya]

Running the sim again myself. Have no trouble getting past ~140s but there's that spike where the cars end up backflipping around 155. A couple managed to get to 158-159 but nothing's managed to get further yet. What I think happens is evolutionary cul-de-sacs. So each person is seeing stagnation kick in at different points, and jumping to a more optimal genetic line is just too much of a jump. You have to mess up the genetic parameters temporarily to break out of the dead-end.

EDIT: current limit = 160.

[scrdest]

150-ish spike is a hard limit for me. I *once* lucked out with a very very pretty machine that resembled a motorcycle frame - low-set, light, dense front wheel, big wheels that cover both sides, shaped like a top-flattened X, very nicely covered terrain - but beating the Backflip Peak is just random chance on each generation, so I'd see regression to essentially random designs which couldn't even get to that point.

That spike is pretty much impossible to consistently beat because the terrain gen derped and so the curves of ascending and descending terrain intersect each other and form a slight backwards curve in each direction; to beat it, you need to land just right to end on top of it, have a frame that won't get stuck on it, and be front-heavy enough to tip over it.

I'm changing the seed to Bay12 now.

[Flying Dice]

I've started running this again.

Code: (As of Gen 160) [Select]

#1: 153.29 d:149.92 h:0/9.69m (gen 137)
#2: 150.55 d:145.34 h:-0.21/7.04m (gen 89)
#3: 149.99 d:144.41 h:-0.21/8.17m (gen 87)
#4: 149.73 d:145.38 h:0/7.31m (gen 116)
#5: 149.71 d:144.65 h:-0.21/7.23m (gen 113)
#6: 149.41 d:145.79 h:0/7.29m (gen 131)
#7: 148.45 d:143.91 h:0/7.44m (gen 115)
#8: 148.17 d:143.91 h:0/7.42m (gen 136)
#9: 148.07 d:144.12 h:0/7.3m (gen 117)
#10: 147.29 d:144.83 h:0/7.98m (gen 138)
The 103-106 triple peak is the main place that weeds out machines, it's practically designed to present three very different challenges in short succession, which makes it unlikely to get anything past there. The 150 peak is, as scrdest said, basically RNG bait: even if you get a car that can beat it, chances are it won't happen again, that car will fail on the next run, and some new piece of crap will replace it, locking you back into trying to get past 103-106.

e: changed seed to Bay12 myself. First run made it to 124m. Second one made it to the 153 cliff. I think I remember trying this seed before and that cliff being a hard limit. It's already really steep and really tall, and the section just prior is perfectly structured to sap momentum.

[Arx]

Spoiler (click to show/hide)

I will not forsake the bay12 world!


Top Scores:
#1: 198.75 d:195.45 h:-8.06/9.36m (gen 1609)
#2: 194.42 d:190.65 h:-8.25/8.43m (gen 1743)
#3: 194.37 d:190.97 h:-7.85/9.28m (gen 1318)
#4: 194.27 d:191.24 h:-7.98/8.61m (gen 1642)
#5: 194.19 d:191.03 h:-7.89/7.98m (gen 770)
#6: 194.01 d:190.23 h:-7.79/9.24m (gen 1645)
#7: 193.96 d:190.43 h:-7.85/9.42m (gen 1527)
#8: 193.93 d:190.92 h:-8.19/8.32m (gen 1638)
#9: 193.91 d:190.96 h:-7.88/9.35m (gen 1499)
#10: 193.86 d:190.88 h:-8.19/8.73m (gen 1620)

Edit:

Well, it's been over 600 generations and nothing's been able to pop the cap. Gonna call it.

I think I decided it was physically impossible for a car to beat 200m, and then gave up.