This topic is for the purpose of discussing #4550: Jigwally's NES Othello in 00:26.42

I'd say it's vaultable. I can dig this, it's really fun to pull off for real.

A quick Google search shows that there are 57 possible setups for a 13-0 win. I don’t believe this to be a TAS optimizable bot-less… Edit: I just noticed there are only 25 rerecords… no way you could’ve even tested the fastest method of movement with so few.

I don't think the 57 possibilities were ever published, but in any case here's another solution with the same number of half-moves:

-- -- -- -- -- -- -- --
-- -- -- -- -- -- -- --
-- -- -- -- 03 -- -- -- 
-- -- -- WW BB 02 -- -- 
-- -- 09 BB WW 08 05 --
-- -- -- 06 01 04 -- --
-- -- -- -- 07 -- -- --
-- -- -- -- -- -- -- -- 

Also, because of the way this game is designed, the number of flips is important to minimize too because of the flip animation. In this movie, it occurs 14 times. I know that at least be 10 flips must occur to finish the game in 9 half-moves, but what about the 4 others? In any case, this may be the fastest route, but it has yet to be proven. As it is now, it's no better than just knowing the series of moves itself.

Hi, thanks for bringing that up. I went ahead and tested a few other possible games. I used these solutions as the basis: https://www.youtube.com/watch?v=6ehiWOSp_wk http://reversi-nxn.blogspot.com/2008/08/shortest-reversi-game-on-boards-7x7-and.html For each of these I also tried the same solutions rotated each of four ways (i.e. the same games but starting with each of the possible first moves of black). For the above two solutions they eventually broke down when the AI refused to do the next possible move. I also tried rotating the game played in my current TAS. Note that in addition to four possible opening moves, on my next turn I have the option of two moves that are mirror images of each other so there are 8 possible iterations to try. Of these 7 additional iterations I was only able to get the AI to make the necessary moves in 2 of them, and those 2 games were slower than my current TAS. One of them was only 4 frames slower though, so it seems very promising that there is an improvement possible. I'll keep researching this with more possible solutions. I can't find a complete list of shortest Othello games but presumably the total count of 57 possibilities includes rotated/flipped forms of the same games.

The rules state that board games are not considered to be "serious games" or eligible for Vault... but I think that's kind of a lousy, over-generalized rule. I don't know what's up with the re-record count or if this is an optimized movie. I only know I appreciate the concept and that I was in no way entertained by the run, so... here's a No vote, with a glimmer of hope that the run might be publishable at all.

From a technical standpoint these sort of games can be pretty interesting, as it's still quite a step from the theoretical shortest game to an actual RNG-manipulated TAS. Then there's a game's "movement" system, which could also make a theoretical perfect solution not the fastest to execute. I did somewhat enjoy this, so a meh vote from me. And a rerecord count of 25 does not have to mean anything, this movie file could very well be the result of thousands of rerecords, that just weren't done with this file loaded.

I'm somewhat surprised by the nearly negative reaction in most comments. If this had drug on for 5 minutes, this would clearly be a no vote, but the length is definitely appropriate for this kind of game. Furthermore, while the rerecord is a bit odd it's not unheard of to do a lot of testing and know exactly what you need to do when you make the final movie, especially if it's being driven by an external script; I see nothing in the movie that looks particularly unoptimal to imply that the TAS was created in a sloppy way. Yes vote for entertainment - it may be a board game, but it's a board game in which the author has soundly abused the AI, which always makes me smile. I may, possibly, try to console-verify this game. No promises. :)

This movie synchronizes on a real console! (using a 1.5ms frame window) Link to video I did some additional testing on the real console and confirmed that there is no way to make the computer think faster after it is the computer's move; all you can do is optimize the cursor movement and placement of your own pieces ahead of time to influence what the AI does. I had a (somewhat depressing) conversation on IRC in #tasvideos where everyone seemed to be of the opinion that, because this is a board game, it is somehow ineligible to be published regardless of the tier. Now, I'm classically a bit of an inclusionist but really, having a prejudice against a well-played game simply because of its genera seems wrong to me. Let me say this: Othello is one of the very few games I owned as a kid and I've perhaps played it more than most people did simply by virtue of limited access to other titles. When I see the AI in this game so artfully manipulated into making bad moves, with such little computer time spent thinking, I come away impressed. To not publish this run on the basis of prejudiced rules that pre-date the creation of the Vault tier strikes me as wrong at every level. Sure, there are probably only a handful of other people on the whole planet that sank a large amount of time into this game and would be impressed by what the author accomplished, but why deny them the opportunity? If the run is deemed unentertaining (a potentially likely view by those who never had to suffer at the hands of this cruel AI) and the run is subsequently only in consideration for the Vault tier, what would be the harm in publishing it? For that matter, why *not* publish Chessmaster in the Vault tier? I'll go back to my corner and try to be stoic about whatever happens next but I personally feel we do a disservice every time we reject a Vaultable movie that may appeal to a small group of people who might have nostalgia for it.

I fully agree with dwangoAC, that we shouldn't discriminate simply based on genre. We have prime examples of board games TASes that have been accepted in Clue and Monopoly. Both required some luck manipulation (Monopoly had lots, Clue, not so much), but aside from that, not much else. Still, I found them very entertaining. I like this type of run. It's short, thoroughly trounces the AI, and I see nothing sloppy or unoptimized about it. The only thing that's keeping me from voting 'Yes' is that it seems it was done on the easiest level of difficulty. Jigwally, can you explain why you didn't choose Level 4? What's the fastest you could do this on that level?

Jigwally, can you explain why you didn't choose Level 4?

Speaking out of turn, but when you do that, you can expect the computer to take a very, very, very long time to make moves. It's downright obnoxious. Still, if it is possible to trick the computer into the same stupid moves that might be interesting (it'd just be a very boring movie).

I'm on board with dwangoAC, here. As for publishing this, why not? We've got the Vault, and it's also a decent way to spend 30 seconds watching a tool-assisted run of a board game getting blitzed through.

I owned this game. I only beat the easy AI once. I have no idea whether it's because I suck at Othello or because I sucked even more back then (many Othello noobs will try to flip many pieces on each turn, which versus a good player will lead to you having very few moves and them eventually forcing you to give up a corner and subsequently a crapton of secure territory). But this TAS (and the real-time speedrun by the same author!!!) entertained me. YES!

Speaking out of turn, but when you do that, you can expect the computer to take a very, very, very long time to make moves. It's downright obnoxious. Still, if it is possible to trick the computer into the same stupid moves that might be interesting (it'd just be a very boring movie).

Okay, that's enough for me to remove my reservation. Yes vote!

Here's a page with more info on the shortest games. Choose the "Wipeouts" section: http://www.samsoft.org.uk/reversi/default.asp It mentions 57 possibilities, and says, "Ignoring reflections & rotations there are 5 with the perpendicular opening, 23 with the diagonal and 29 with the parallel." The left-hand list has 24 entries, but many of those have variations linked in the description, so I think all of the unique cases are covered. Now all that's left is for someone to count the flips and find promising candidates for improvement ;)

I don't see any problem in publishing this kind of TAS. One question, though: Are there difficulty levels, and if so, was this played on the hardest?

I fully agree with Scepheo's post. That post mentioned pretty much everything I was going to comment on, like the difference between the shortest possible game and the actual TASing of said game. I don't know if I found this particularly entertaining to watch, but it wasn't boring because it was short. It is funny to see the CPU take such a dumb strategy right off the bat. I think I'll vote Yes.

Here's a page with more info on the shortest games. Choose the "Wipeouts" section: http://www.samsoft.org.uk/reversi/default.asp It mentions 57 possibilities, and says, "Ignoring reflections & rotations there are 5 with the perpendicular opening, 23 with the diagonal and 29 with the parallel." The left-hand list has 24 entries, but many of those have variations linked in the description, so I think all of the unique cases are covered. Now all that's left is for someone to count the flips and find promising candidates for improvement ;)

Thank you so much for finding that. I checked and perpendicular 5 and every diagonal/parallel solution is tied for 14 flips. So I guess the next step is to assign a value for how long cursor movement takes for any particular square and find the lowest value that way.

I also think there shouldn't be total disqualification for board game genre for Vault. And I also had this game when I was 8. I hated it, but I saw my relatives play it A LOT, so I like how fast it was beaten here. Voting Meh.

Personally, I wouldn't count this as beating the game because it doesn't reach the credits. If you beat level 4, you're then given a match vs the cpu at level 5. Only if you win that do you get the credits with the programmers' names.

How slow would beating it be on the hardest level? I think that would be interesting even if it were slower, to come up with the fastest successful strategy. It doesn't seem to me that it would take that long.

It doesn't seem to me that it would take that long.

You have no idea how many minutes it can take on just a single turn until you've tried it. (It might not be literal minutes, but trust me, on the harder difficulty levels it seems like the computer is trying to take every bit of its allotted time). I would consider getting to the credits to be a different category; this run is akin to "fastest win" and the other is more of an any%. However, if this run is rejected based on the existing vault rules, and a new run was created which reached the credits, I assume it would have a higher chance of making it through the gauntlet of judging based on the existing description of the Vault tier. I suspect it would be a far more boring run, though...

I liked it. I think it's worth publishing :) Definitely helps that it's short so I don't mind throwing out the idea of beating level 4 and 5 since it would up the thinking time. Worth looking at in the future, but not at the expense of publishing this run.

I don't see any problem in publishing this kind of TAS. One question, though: Are there difficulty levels, and if so, was this played on the hardest?

This was done on the easiest difficulty. As with most board games in this era, the high levels take a really really long time to move. And generally the easiest level involves some randomness as that is a quick easy way to lower the AI's skill level. Thus, hardest difficulty means a very long movie of watching the computer think, and the easiest involves a trivial manipulation of the RNG to make the CPU make a horrible blunder. See Battle Chess, and Chessmaster TASes and discussions as this is pretty much the same situation.