GonZaLO has chosen to double down on with being rude and a bad sport. He's shown his true colors; so be it. This confirms my earlier impression of him, when playing the TRL Season 29 playoffs; he was gloating quite smugly in the chat, showing himself to be a bad sport when winning. And now he has shown himself to be a bad sport by way of denying he had a high likelihood of losing.
I withdraw my proposal to continue this game live online (in the case that I could have implemented it in a build soon enough); it was a bad idea anyway, as it would require either giving him a private build that he doesn't deserve, or publically releasing a build that would essentially be scaled-down TA build. Implementing the replay-continuation-online would be easy, but a week or two is not anywhere near enough time to implement and debug the security/protection against abuse of the replay-continuation feature. I do however plan to try implementing recovery-from-disconnection in time for the Season 30 TRL playoffs. This would be done by waiting for and allowing disconnected players to reconnect, instead of resuming from the replay file, to avoid potential abuse. (As such, it would only be able to recover from disconnections, not power failures or crashes.)
[Edit: I neglected to calculate the chances for draws when writing this last night. Now added. So as you can see, GonZaLO's wild guess of being 60% likely to pull a draw is not even close.]
FWIW, I made the replay
2022-05-01 04.40.21 {1 - non-TA possible play @ 12m53s} 2022-04-30 23.52.02 [Online Round 2] @Deadcode, GonZaLO.WAgame during a night when I had a headache and slight light-headedness from the stale air caused by keeping my windows closed for many hours due to there being lots of nasty campfire smoke outside. When having played the TRLs with GonZaLO earlier that morning, however, I was in reasonably good condition breathing fresh air (and I make a point of only playing ranked games in those conditions). So this replay (and the whole session in which I analyzed it a bit further) doesn't really represent my best mental effort.
The probability of the Armageddon in that replay (done at 23:08.46) resulting a win for blue was about
13.8%, with a
11.5% chance of a draw. If blue delayed using it until a later turn, or took better previous turns, the probability would increase.
Senator, thanks for doing your tests. Here are the approximate probabilities of the Armageddons you did winning the game for blue:
23.0% blue win (
9.65% draw) at 3:00.34 in
2022-05-01 11.56.33 [Offline Round 2] DC vs GonZaLO 1.WAgame17.5% blue win (
11.5% draw) at 3:51.08 in
2022-05-01 12.04.51 [Offline Round 2] DC vs GonZaLO 2.WAgame11.3% blue win (
5.34% draw) at 4:10.80 in
2022-05-01 12.10.46 [Offline Round 2] DC vs GonZaLO 3.WAgame - better than in my previous posted replay
22.5% blue win (
4.89% draw) at 3:18.94 in
2022-05-01 12.25.02 [Offline] DC vs GonZaLO 4.WAgameThis seems to show that probably only the plays in your third replay were potentially more optimal for red. There is also the possibility in the first replay for blue to delay using the Armageddon, potentially increasing the chance of it working, by roping over and Girder-blocking the other worm.
But these were not a true continuation of the game but rather an approximated mockup. The presence of crates, oildrums, etc., changes things. I recreated your third replay (with a slight modification – my worm on the left places a Girder instead of just skipping go). Note that you were off by a turn of SD (as well as other differences).
20.6% blue win (
11.5% draw) at 21:48.48 in
2022-05-03 04.28.35 {8 - possible play (based on Senator's) @ 12m53s} 2022-04-30 23.52.02 [Online Round 2] @Deadcode, GonZaLO.WAgame
(136.62 kB - downloaded 14 times.)
So, apparently it was less optimal play than mine. (Note that the Armageddon in this replay file shows a win for blue, i.e. the thing with a 20.6% chance happened, which contradicts GonZaLO's claim that I'd never post a replay where he wins.)
Anyway, I've now analyzed this endgame in depth. Here is the optimized play I came up with:
2022-05-03 02.38.49 {possible endgame @ 12m53s} 2022-04-30 23.52.02 [Online Round 2] @Deadcode, GonZaLO.WAgame
(150.28 kB - downloaded 9 times.)
There were four opportunities to use the Armageddon. The approximate probabilities of that resulting in a win for blue:
0.92% blue win (
0.13% draw) at 20:01.62
0.58% blue win (
0.00% draw) at 20:45.80
1.65% blue win (
0.05% draw) at 21:50.16
15.4% blue win (
6.38% draw) at 22:55.62 (last and best chance)
Here is a replay of the 15.4% chance happening (a win for blue):
2022-05-03 04.49.44 {possible endgame @ 12m53s} 2022-04-30 23.52.02 [Online Round 2] @Deadcode, GonZaLO.WAgame
(144.43 kB - downloaded 7 times.)
This is indeed more optimal for red than my original replay. Blue could have done some better moves in my original replay, resulting in a
21.4% chance of winning (
8.03% chance of a draw) from Armageddon:
2022-05-03 05.39.10 {possible play @ 12m53s} 2022-04-30 23.52.02 [Online Round 2] @Deadcode, GonZaLO.WAgame
(148.72 kB - downloaded 6 times.)
So I guess Senator's is marginally better at 20.6% (but still worse than the 15.4%).
On the other hand, red could have done some better turns too in a variation of my original replay, especially with some decent wind – here is a scenario even more optimal for red than the above (assuming blue can't do something much better under these circumstances):
2022-05-03 06.51.23 {possible endgame @ 12m53s} 2022-04-30 23.52.02 [Online Round 2] @Deadcode, GonZaLO.WAgame
(155.97 kB - downloaded 13 times.)
5.98% blue win (
0.09% draw) at 19:54.96
4.59% blue win (
0.09% draw) at 20:41.70
2.78% blue win (
0.22% draw) at 22:19.30
5.09% blue win (
1.20% draw) at 22:49.86
8.25% blue win (
7.00% draw) at 24:07.66 - the replay file shows this 8.25% chance win for blue happening
12.1% blue win (
3.66% draw) at 24:07.66 - if traversing to the left side (wind allowing) to use the Armageddon:
2022-05-03 18.57.38 {possible endgame @ 12m53s} 2022-04-30 23.52.02 [Online Round 2] @Deadcode, GonZaLO.WAgameI'm guessing, from GonZaLO's tone, that he may think I tend to make gameplay choices that he considers so dumb he'd never make them himself. But if so, he is seeing that through the lens of his own playing style, ignoring that I probably make many other types of decisions, and types of moves, better than he would. It likely balances out much more than he apparently thinks, but I guess we'll never know because GonZaLO refuses to play a statistically sound number of games with me, choosing instead to rest on his laurels that are quite possibly just a result of being lucky.
And for that matter, we may never know how good he really is when playing against other players, because he has the unfair advantage of disconnecting often, not only voiding those games but concealing them from the public by reporting only the non-disconnected replays. Even if he upgrades to a better ISP, it would take asking each player individually to see if he's really stopped having lots of disconnections, because he refuses to let his disconnected games be a matter of public record. (Adding recovery-from-disconnection to WA would of course fix this, but that will be too late for the bulk of this season of TRL T17, and possibly too late even for the playoffs.)