I just joined Speedrun.com, and I am attempting a speedrun of this game. I thought I would share what I've found out about the CPU crush strategies. Here's the crush score or situation that will normally cause each opponent to use the attack:
Spider: 10 Bat: 10 Skeleton: 20 Treasure Chest: 20 Ghost: 20 Snake: Will always wait for you to use it first, then retaliate Zombie: 20 unless it has more than 3 crush bars, otherwise 10 Scorpion: 30 unless it has more than 4 crush bars, otherwise 10 Witch: 10 Mummy: 20 unless it's in danger of losing, otherwise 10 Sphinx: 20 unless it's in danger of losing, otherwise 10, and it will intentionally target your magic gems!
I'm not sure about the rules for the skeleton, treasure chest and ghost for switching to using it every chance they get. Still, this is useful for knowing when using your attack will cause your opponent to immediately retaliate - the most glaring one besides the snake is the scorpion exceeding 4 crush bars with an attack. So when it's at "crush tolerance", you should not use your attack when you have a useful piece, as SegaJunkie explains in his guide to speedrunning this game.
I'm doing this on a Fusion emulator which has built-in video recording. However, I must admit that it's easy to pause emulation and not affect the appearance of the video - I've tested it. So you could theoretically stop and think about your next move, but the advantage this would give a player good enough to beat the sphinx at all is small. If anything, it would be detrimental to me, so even if I were morally willing to cheat like this, I would still be better off doing it for real. So it may not be worth worrying about.
Standards of proof for this game might have to be a little lower because this game only has a few players, and it's so hard to get a time at all for Hard mode! In fact, they do seem to be in comparison to more popular speedgames - the American record, and the only one in the USA, does not have video proof, but does have a witness.
I'm tantalizingly close to doing this. A playthrough of mine where I simply tracked play times for each game came VERY close to what would have been the American record after adding in the mandatory delays in the game, which is almost 30 seconds per two out of three match, assuming the first two games are won! (I recently edited this post to indicate better what what the zombie does.)
Thanks for the breakdown on this, however I must point out that there is a difference between both versions of Columns 3 (Japanese and English) Sphinx will knock out monocolors in the English version opposed to the Japanese version where Sphinx will not knock them out.
Thanks for the additional info, @RoyTheDragon! I'll see which version I have by seeing if the sphinx knocks out one of my pieces with three of the same color before reaching 20 while at or below four crush bars (like the scorpion, and I recently edited the zombie part of my post to say a 3 bar tolerance instead of 2) and not in danger of losing, which is what I assume you meant about monocolors, how I'm playing it. I'm actually going to shout out to a 1994, arcade-only game called Stack Columns which you actually seem to have the world top that's been proven on video, at 360,391 while "1 credit clearing" the game and actually getting the truth behind your father's apparent death in that game. And since this is a "quickplay", even though it lets the big time bonus run out at the end, it still makes for a worthwhile speedrun time on this site to me. I've 1CC'd it once, with 214,231, with an attempted video recording but the video was pretty much lost a bit into the run, and it was originally intended only as a test too!
Not gonna lie, I'm pretty much considered a GOAT of the series by many although I don't really consider myself one to be quite honest.
I'm just getting interested in speedruning this game and I wanted to know if Japanese or English is faster and which one is easier to speedrun
the Japanese one is easier to run because the Sphinx does not shatter when you are dropping 3 of the same color and will not pressure you when you're up high.
