Grobots - Tournaments - 2002

[tournaments from 2003 and 2004]

Big Disorderly Tournament

(December 21, 2002) We ran nineteen sides overnight, for about 100 rounds per side. We used the current default rules (10x10 world, 10 sides per round, 5000-100 seed.) Gnats was excluded because of its occasional crashes. Productive Plus is a modified Productive. Sugar Pixies and Decoys are unfinished experiments.

RankSideAuthor ScoreSurvivalEarly
Food sources (%)
1Intoxicated 3Daniel von Fange 28.7%35%37%5148
2World Toad 3Devon 16.4%29%53%255320
3Commune 2Devon 15.9%42%18%52415
4Business Cycle 2fMatt Burkholder 15.2%33%31%69254
5Productive 5fDaniel von Fange 14.9%24%59%100
6Productive PlusDaniel von Fange 14.1%20%58%100
7Eventually 12Devon 11.3%17%63%100
8Microb 2Matt Burkholder 11.1%48%10%37556
Not Quite MadWarren 11.1%36%19%5868
10Teledont 6fMatt Burkholder 10.9%22%52%136124
11ActiveWarren 10.3%28%29%7920
12Not Quite WiseWarren 8.5%36%28%48510
13Four Winds 2Warren 7.2%27%32%8118
14Gunner 2Warren 6.3%19%46%100
15Grudge 2Devon 4.4%11%48%100
16Ants 2Warren 3.8%31%31%46457
17Sugar PixiesDevon 1.6%15%38%7722
18Missile-lunatic 4Warren 0.9%10%49%100
19DecoysDevon 0.2%9%50%8118

Yet Another Tournament

(November 30, 2002) 51-69 rounds per side. Lunatic snuck in somehow, and beat Gnats. Check out the new improved Teledont.

RankSideAuthor ScoreSurvivalEarly death rate
1Intoxicated 3Daniel von Fange 32.1%39%16%
2Eventually 12Devon 26.0%30%50%
3Teledont 6Matt Burkholder 14.5%30%39%
4Productive 5Daniel von Fange 11.3%18%54%
5World Toad 3Devon 11.1%26%29%
6CommuneDevon 8.9%27%23%
7Grudge 2Devon 8.2%19%46%
8Microb 2Matt Burkholder 7.4%45%6%
9Four WindsWarren 6.8%27%22%
10Not Quite WiseWarren 6.4%35%17%
11Gunner 2Warren 5.7%17%56%
12Lunatic 2Devon 4.4%16%26%
13Missile-lunatic 4Warren 3.9%13%44%
14Gnats 8Matt Burkholder 2.1%9%53%
15Circle the WagonsMatt Burkholder 1.5%10%31%

Crime Pays

(November 20, 2002) Now that cooling cost starts at 0, Teledont is unseedable. I added some new sides (Commune, Ring of Fire) and some old ones (Poison Ivy, Missile-lunatic), and upgraded Megadont and World Toad. Because there are more than ten sides, the scores add up to more than 100%, and the number of rounds varies slightly by side (it's around 25).

RankSideAuthor ScoreSurvivalEarly death rate
1Intoxicated 2Daniel von Fange 32%42%23%
2World Toad 3Devon 21%43%37%
3Gunner 2Warren 14%36%43%
4CommuneDevon 13%42%38%
5Eventually 12Devon 11%17%55%
6Productive 5Daniel von Fange 9.4%13%47%
7GrudgeDevon 8.4%22%44%
8Microb 2Matt Burkholder 7.4%43%11%
9Missile-lunatic 4Warren 6.7%18%33%
10Gnats 8Matt Burkholder 5.6%19%44%
11Megadont fixedMatt Burkholder 4.0%26%56%
12Ring of FireDevon 3.9%10%45%
13Iron BubbleDevon 1.9%10%29%
14Poison Ivy 4Warren 1.4%9%75%

The effects of a rule change

(November 16, 2002) With automated tournaments, it's now feasible to run experiments. This tournament has the same sides as the previous one, but was run with quadratic cooling cost. Some of the difference is noise, but the effect is clear. In the previous tournament, there are giants in four of the top five sides; in this one there are only two. 30 rounds:

RankSideAuthor ScoreSurvivalEarly death rate
1Eventually 12Devon 32%37%60%
2Gnats 8Matt Burkholder 15%33%17%
3Productive 5Daniel von Fange 14%23%43%
4World Toad 2Devon 13%23%33%
5GrudgeDevon 7.4%17%60%
6Gunner 2Warren 7.0%17%57%
7IntoxicatedDaniel von Fange 4.9%7%53%
8Microb 2Matt Burkholder 3.4%23%17%
9Teledont 5Matt Burkholder 2.2%7%77%
10Iron BubbleDevon 1.1%7%50%


(November 16, 2002) 20 rounds with the new automated tournaments. It didn't take terribly long, and I was doing other things meanwhile. New Eventually and Grudge, which did much worse than I expected.

RankSideAuthor ScoreSurvivalEarly death rate
1Productive 5Daniel von Fange 20%25%60%
2Eventually 12Devon 16%20%65%
3IntoxicatedDaniel von Fange 15%25%35%
4World Toad 2Devon 14%35%35%
5Iron BubbleDevon 7.9%25%25%
6Gunner 2Warren 6.9%20%55%
7Gnats 8Matt Burkholder 6.8%20%50%
8Teledont 5Matt Burkholder 6.2%10%80%
9Microb 2Matt Burkholder 3.3%15%10%
GrudgeDevon 3.3%10%80%


(November 13, 2002) New Eventually and Iron Bubble. 21 rounds:

RankSideAuthor ScoreSurvivalEarly death rateScore when doesn't die earlyScore when survives roundComments
1World Toad 2Devon 24%43%30%36%55%
2Productive 5Daniel von Fange 22%24%45%42%92%
3Gnats 8Matt Burkholder 17%24%45%32%70%Three 100% wins.
4Eventually 11Devon 14%14%62%38%100%Now delays building sentinels, for growth and nonagression.
5Gunner 2Warren 5.6%10%75%24%59%
6Iron BubbleDevon 4.4%10%45%8%46%Very simple armored autotroph.
7Microb 2Matt Burkholder 3.6%29%10%4%12%
8Teledont 5Matt Burkholder 3.5%10%65%11%37%
9IntoxicatedDaniel von Fange 2.9%5%40%5%60%
10Missile-lunatic 4Warren 2.8%10%75%12%30%

Early deaths were just under 50%, and overall survival was 18% (25% in non-elimination rounds), because 48% of rounds were won by elimination. I think most sides pay too much attention to offense and not enough to defense. Iron Bubble, with heavy armor and no offense, beat Teledont and Intoxicated, two sophisticated agressive sides.

Strategy kills

(November 11, 2002) New Gnats, Teledont, and Eventually. No more Fighters. 30 rounds:

RankSideAuthor ScoreSurvivalEarly death rateScore when doesn't die earlyScore when survives roundComments
1World Toad 2Devon 40%63%10%44%63%Good defense. Doesn't start fights.
2Productive 5Daniel von Fange 12%20%63%33%60%Corner hiding matters more than missiles.
3IntoxicatedDaniel von Fange 8.9%17%57%21%53%
4Gnats 8Matt Burkholder 7.7%23%43%14%33%
5Gunner 2Warren 7.6%20%67%23%38%Good defense but starts fights.
6Missile-lunatic 4Warren 5.8%17%60%14%35%
7Microb 2Matt Burkholder 5.1%17%10%5.7%31%
8Teledont 5Matt Burkholder 4.5%10%73%17%45%
9Eventually 10Devon 4.2%10%70%14%42%Now with smarter missiles, but doesn't live to use them.
10Poison Ivy 4Warren 3.5%13%80%18%26%

Overall survival was a dismal 21%, early death rate was 53%, and 30% of rounds were won by elimination (mostly by World Toad). Eventually is in ninth place, and it's not because this version is any weaker than its predecessors. With the new Gnats, there are now two sides that retaliate effectively when shot at. This means sides that shoot at everything they see - Poison Ivy, Teledont, Eventually, and Gunner - get killed. Those four have the four highest early death rates, higher than Productive or Intoxicated. I'm glad to see strategy being decisive.

Microb's small cells are intended to be too cheap to be efficiently killed with missiles, and it works. In one round, it held off all three missile users (and got Productive to kill itself) and scored 78%. I've also seen it defeat six World Toads the same way. The shields on the Rat-derived type make a big difference. Microb has a splendid early death rate. There should be lots of interesting variations on this theme.

My confidence in tournaments falls again

(November 5, 2002) Replaced Homesick with Intoxicated. 20 rounds:

RankSideAuthor ScoreSurvivalEarly death rateScore when doesn't die earlyScore when survives roundComments
1Eventually 9Devon 21%30%65%61%71%I don't know why the early death rate is so high.
2Gunner 2Warren 18%35%55%41%53%Good defense.
3World Toad 2Devon 18%50%30%25%35%
4IntoxicatedDaniel von Fange 12%15%50%24%80%Affects balance of other sides.
5Teledont 4Matt Burkholder 11%35%30%16%32%
6Productive 5Daniel von Fange 7.8%20%65%22%39%Often killed by Intoxicated.
7Microb 2Matt Burkholder 3%30%20%4%10%
8Missile-lunatic 4Warren 2.8%20%40%5%14%
9Poison Ivy 4Warren 2.7%15%60%7%18%
10Fighters 5Devon 1.7%5%40%3%34%

Presumably because of Intoxicated, overall survival is down to 26% and early death rate is up to 46%. 25% of rounds were won by elimination.

I'm glad to see Fighters in last place, but not to see Gunner in second. They are both ancient sides that shouldn't have a chance against good opponents. Intoxicated's success indicates that sides need more early defense, which is probably why Gunner did so well.

Individual early death rates

(October 31, 2002) Same sides as before, but with Eventually 8, which (as you can see) is much better than 7. 20 rounds:

RankSideAuthorScoreSurvivalEarly death rateScore when doesn't die earlyScore when survives round
1Eventually 8Devon31%40%35%47%77%
2World Toad 2Devon12%45%10%13%26%
3Productive 5Daniel von Fange11%20%65%32%55%
4Gunner 2Warren11%45%35%17%24%
5Fighters 5Devon10%25%20%12%39%
6Poison Ivy 4Warren9%35%25%13%27%
7Missile-lunatic 4Warren4.8%25%40%8.1%19%
8Teledont 4Matt Burkholder4.6%30%35%7.1%16%
9Microb 2Matt Burkholder4.3%50%5%4.5%8.6%
10Homesick 3Warren1.5%10%40%2.5%15%

Overall survival was 32% (more than three sides per round) and early death rate was 31%, so the average score for a side that didn't die early was 14%, and that for survivors was 31%. 25% of rounds were won by elimination.

Now counting early deaths

(October 30, 2002) I replaced Business Cycle with the repaired Teledont, and upgraded a few others.

RankSideAuthorScoreSurvivalFractionComments and reasons
1Productive 5Daniel von Fange31%45%69%Hiding in corners really helps.
2Poison Ivy 4Warren21%50%42%Impervious to missiles.
3Eventually 7Devon14%25%56%Recent upgrades made it worse.
4World Toad 2Devon14%55%25%Excellent survivor.
5Gunner 2Warren9.25%30%31%Good defense, even against missiles.
6Homesick 3Warren3.75%25%15%
7Fighters 5Devon2%10%20%
8Microb 2Matt Burkholder1.3%30%4.3%Runs, but can't hide.
9Missile-lunatic 4Warren1.2%15%8%
10Teledont 4Matt Burkholder1.1%20%5.5%Dies.

Overall survival was 30% (yay). Only two rounds were won by elimination (both by Productive). In the 15 of 20 rounds in which I remembered to check, 35% of sides died before 4500 frames, and another 35% died later. Eventually, Productive, and Teledont had high early death rates.


(October 26, 2002) I replaced Algae and Fool with Fighters and Missile-Lunatic. Fighters did much better than I expected. Overall survival was 25%. 3 of 15 rounds were won by elimination.

1Eventually 6Devon28%40%69%
2Productive 5Daniel von Fange26%33%77%
3World Toad 2Devon20%60%34%
4Fighters 4Devon13%40%34%
5Poison Ivy 4Warren4.5%27%17%
6Homesick 3Warren3.9%13%30%
7Gunner 2Warren2.8%13%21%
8MicrobMatt Burkholder0.3%7%5%
9Missile-lunatic 3Warren0.1%7%2%
10Business CycleMatt Burkholder0.07%7%1%

32% survival!

(October 24, 2002) Missiles have been weakened, and it has improved the feel of the game. Missile-users still took the top two places, but they appear to be beatable now. Only two of 15 rounds were won by elimination. Overall survival is up to 32%. Early deaths are under 30%, and would be even lower if Productive didn't blunder and die so often.

1Eventually... 5Devon38%60%63%
2Productive 5Daniel von Fange18%27%68%
3Gunner 2Warren15%60%26%
4World Toad 2Devon12%60%20%
5Poison Ivy 4Warren9.7%40%24%
6MicrobMatt Burkholder2.7%40%6.7%
7Homesick 3Warren2.6%27%9.8%
8Business CycleMatt Burkholder0.3%7%4%

I am still tired

(October 3, 2002) 16 rounds:

1Productive 4Daniel von Fange38%38%100%
2Eventually 5Devon19%31%59%
3World Toad 2Devon17%44%38%
4Fighters 3Devon10%25%41%
5Gunner 2Warren7.5%19%40%
6Poison Ivy 2Warren4.8%6.3%76%
7Teledont 2Matt Burkholder1.9%6.3%31%
8Missile-lunatic 3Warren0.2%6.3%3%
9Life and Death 4Warren0.06%13%0.5%
Rat without shields 2Warren

Algae was only in four rounds, and Rat-no-shields in twelve. Productive didn't have radio hardware yet, so it formed a blob instead of a line. It seems to work fine anyway.

Overall survival was 19%; survival in non-elimination rounds (8 of 16) was 27.5%.

I am tired

(September 28, 2002) Look who won.

1Productive 3Daniel von Fange53%55%97%
2Eventually 3Devon15%15%100%
3World ToadDevon12%25%49%
4Poison Ivy 2Warren8.7%15%58%
6SunflowerMatt Burkholder2.5%5%50%
7Gunner 2Warren1.7%10%17%
8Missile-Lunatic 3Warren1.5%15%10%
9Teledont 2Matt Burkholder0.1%5%2%
10Rat without shields 2Warrenabout 05%about 0

Of 20 rounds, 13 were won by elimination. Overall survival was 16%; survival in non-elimination rounds was 27%.

Missiles are too powerful

(September 25, 2002) These results speak for themselves. I ran 14 rounds, and Eventually won ten of them by elimination.

1Eventually 3Devon71%71%100%
2Poison Ivy 2Warren18%21%83%
3Teledont 2Matt Burkholder7.3%14%51%
4World Toad (old version)Devon1.8%14%13%
5Gunner 2Warren1.4%7%20%
6Missile-Lunatic 3Warren0.07%7%1%
7Flyswatter 2Warren00
Life and Death 4Warren
SunflowerMatt Burkholder

The fraction of sides which die early is down to something like 40%. This is why Eventually is doing better - whenever it survives long enough to build missiles, it wins. The early death rate is down because sides are getting better at defense, so overly agressive sides die before they can kill many others. Teledont in particular is no longer surviving well.

Short tournaments are unreliable

(September 23, 2002) Results of today's tournament:

RankSideAuthorScoreSurvivalFirst HalfSecond Half
1Teledont 2Matt Burkholder32%40%16%48%
2Eventually... 2Devon30%30%40%20%
3Poison Ivy 2Warren12%15%8.2%16%
4Life and Death 4Warren10%20%19%0.9%
5Gunner 2Warren7.3%25%9.5%5.1%
6SunflowerMatt Burkholder4.1%15%2.7%5.5%
7Flyswatter 2Warren2.1%15%0.2%4.0%
8Missile-Lunatic 3Warren1.9%10%3.8%0.0%
9Cycle 3Devon0000
Life and Death 3Warren

I included two versions of Life and Death to see if 4 really was an improvement over 3. I guess so.

The tournament had twenty rounds, but I originally ran just ten. Eventually won, but I thought it might be a fluke, so I ran another ten rounds. The last two columns show the results of the first and second halves. You can see how different they are.

The moral: ten rounds is not enough to overcome noise. This may be partly because present sides tend to fight unpredictable battles. More advanced sides might have less random results.

Overall survival was 17%.

Combined results from a few tournaments

(Sepember 21, 2002) I ran a few little tournaments recently. The top scorers in the combined results:

1Teledont 2Matt Burkholder42%74%19
3Gunner 2Warren16.4%34%29
4Poison Ivy 2Warren15.6%21%29
5Cycle 3Devon9.4%35%20
6Flyswatter 2(?)Warren6.4%33%15
7SunflowerMatt Burkholder4.6%17%18
8Life and Death 4Warren3.2%4%20
9Missile-Lunatic 2Warren2.8%33%24

I included survival scores because they're easy to calculate and they provide useful information. For instance, Missile-Lunatic survives often (because of its missiles and passive dodging?) but scores low because of its low growth rate.

I'm not sure what version of Flyswatter I used.

[This was the first tournament.]

Grobots by Devon Schudy ( and Warren Schudy (