Well this first week of Guild Wars is over and we now have some numbers for AI comparison.
I play on both platforms but primary on console. I’ve ALWAYS been a big proponent of the AI on console for the fun factor, I find the PC AI to rote like. I have posted numerous times about this in the past. The major difference on Unity is that you can not reliably Skull bait the AI. Effectively giving yourself a free turn as the AI pokes you with a Skull.
Now that the First Guild Wars on console is over what surprises me the most is the point differences at the top of Bracket 1 in Guilds Wars leaderboard. On PC/Mobile they are a about 50% higher than console. This is the first time we can almost compare the AI performance numbers with and without Skull baiting, vs my perception of the differences over the last 2 years. The Numbers for the top cluster of Guilds on X1 are 500-700k vs 800-1,000 on the PC. Basically 300k!
Now we still don’t know if they are going keep the console Unity AI when PC/mobile is migrated to Unity or not. If so then the PC/Mobile Guilds might be surprised how they won’t be able to reliability control the AI.
I don’t know what is going to happen in the migration at all. But if the PC gets get the Unity AI I hope we don’t get a rehash of “the AI is out to get me”, the AI is cheating, the AI knows what’s above the board, the AI is doing a Mario cart catch up when I’m winning, etc threads.
Totally agree. Skull baiting is the only way to beat many teams in guild wars as it buys you time to cast first. If it changes we’re going to see a LOT of complaints
The salt would be so thick you wouldn’t be able to read the forums at all. Look how bad a 5% winrate hit from Famine teams made the forums. a 20-30% win rate hit would cause heads to roll i think.
Interesting info on the score difference between platforms. Personally I’m excited about the possibility of the console AI on mobile, although I’m hoping the lack of skull baiting is due to better strategy and not just being totally random. I’m also looking forward to the ability to customize gem match priority (which I know is currently only for PvP, but they’ve promised it for GW “soon”). However, I’m not too keen on what I’ve heard about the match-making, and an inability to nerf your defense in order to get some opponent variety.
As for complaints, regardless of direction, there will still be lots of complaints. Like with almost every change, there will be a bunch who love it, a bunch who hate it, and a bunch who will claim random isn’t random. Should be fun
@beanie42 It’s difficult to describe as we don’t have full access to the AIs Gem matching priority algorithm (nor should we). I’ll give you some detail with the console AI with as little “flavor” as possible.
The AI will take a non-optimal matches with regularity. (like a 4 over a 5 etc.)
The AI will ignore a Skull match occasionally, using its default settings. (best guess 10-20%, just enough to cause Skull bait problems)
The results of these non-optimal gem matches are random (unexpected cascade for AI, setup the human player, no real board effect)
In general terms the AI plays less “efficiently” than on PC. HOWEVER, without knowing its actions next turn with 100% certainty it becomes a more difficult opponent.
The AI slider bar has a large influence over gem priorities. You can help or hurt your Defend team based on your settings.
Thanks for the reply @Strat . The part that concerns me most is that the AI might “ignore it’s own programming”. Specifically, if the default AI has all the colours and skulls at an equal priority, it’s understandable that sometimes skulls would be ignored. And while this might appear to be random, it’s quite explainable.
And the screenshot (thanks for posting it @Rickygervais ) doesn’t contain any numbers, but for the sake of argument, lets assume the scale is from -5 to +5. If colours are set at 0 and skulls are at +5, I would expect the AI to NEVER miss a skull match, similar to the current PC AI. I’m hoping that the devs simply have the “defaults” set to something like maybe 3 which is why it would occasionally miss some, but that if somebody “dials it up” then it would always take skulls.
My reasoning is simply that to me a big part of “real strategy” comes from learning and adapting to your opponent. So after playing a dozen moves or so, if I’m observant/skilled, I should be able to guess what the AI’s next move is likely to be. I should be able to skull bait an AI programmed to favour skulls, but not skull bait one favoured to take a certain colour, and each game/AI should have it’s own “feel”. Hopefully I’m describing that properly, and if so, does that sound close to how it works or is it significantly/noticeably more random than that?
