How often are YOU running into the "super lucky" opponents?

Thanks for proving my point.

The AI in the last CONSOLE update got a small rewrite so that it could hook into the user-settings for mana-preference on defense.
ON PC/Mobile there hasn’t been an AI update/change for many months…
In fact we can’t really update most of the AI without issuing a client update on either version, so any there couldn’t have been a significant change since the previous updates.

As part of that rewrite we tinkered with the priorities of when it would & would NOT select various colors… how much it tried to grief you by taking YOUR colors (we actually toned that down just a little), and it’s chances of taking skulls and aoccasionally passing up a 4/5-of-a-kind. It made it a little more unpredictable, but not really much smarter. It certainly wouldn’t have affected any “luck”.

Now I can vouch for some of the code on PC/Mobile, because I wrote it…
On “Normal” difficulty it cheats in the human’s favor, and makes sure the AI cannot get a “lucky” gem fall (either setting up a cascade of skulls, 4-of-a-kind, etc…) unless the human has already had twice as many of these events.

That same code exists on Consoles, but I am not able to verify if it is actually working. I WILL check with the console team on Monday though, and resond here to let you know.

In no version of the game is there any code that is able to affect the “luck” or “starting board” or anything really, to advantage the AI Player. Except for ONE THING… which I will mention here just for completeness… if ANY team (AI or Human) gets too many turns in a row (more than 5 turns, using a looping team) we decrease the chance of them getting an extra turn to keep the loop going.

9 Likes

Maybe we get tothe point here, i can assure you that AI get lucky cascade without us having any luck before.

Yesterday i did have one of these game. Sylv/valk/just/mab against a goblin team.
I picked a three yellow match, giving me four due to banner, then it was AI turn.
AI filled three troops entirely and the last one was 8/10. With a combinaison of cast, extra turn, lucky cascade, cast, extra turn, lucky cascade, etc, etc, etc; the AI just killed two of my troops. All i was left with was a valk 4/10 and a mab 0/15.

Shortly, making the AI better was great and it works fine. But sometime, it seems the anti cascade AI have is not working.

For more info, this pattern seems tied with pvp games, as it never happens in explore and i do a lot of them.

Hope it helps
Fred

Ultimately I will say that it doesn’t matter if the AI has changed or not; what really matters is player perception and if most players feel like the AI is cheating or extra lucky, then that needs to be addressed by the devs. Luckily, it seems like @Sirrian feels the same way so I’m sure the devs watch for this and will adjust the AI or give players other modes if it this feeling spreads to a significant part of the player base.

I totally agree.
Too many times I feel cheated by the AI.
Too many lucky drops for the AI.
Too many chains of 4x or 5x in-your-face for the AI.
Too much “mana screw” for the player.
I found really difficult to believe that the drop is really totally random. Really really difficult.

On the PC/Mobile version that can only happen in:
a) Ranked PvP, where we give a fair match by not cheating in the human player’s favor
b) If the difficulty is set at hard or higher.
If it is behaving otherwise, it’s a bug, but I have to say, given my history with gem matching games, I believe it is more likely to be an error in observation, rather than in the code (not calling anyone a liar here, but our brains are just really poorly wired to observe this stuff, as many many studies have shown).

Honestly, I would actually LOVE if someone could provide us video evidence showing NORMAL MODE in a NON-RANKED-PVP game where this happens, as if it IS a bug, nothing would make me happier than to fix it !!!

ON console however, as I mentioned, I will check with the console team on Monday.

3 Likes

Sorry. I am complaining about PVP.

What point, exactly? How does my pointing out you’re being obnoxious and sanctimonious while shoving your own opinions and “truths” down people’s throats prove anything you’ve been rambling on about in this thread?

Rhetorical question again - it doesn’t.

1 Like

I would certainly appreciate that. I shall keep that in mind about the part of the code you mentioned, and be sure to check that against my findings in the future if/when I run into overly frustrating opponents again that, to me, appear to be cheating.

Despite my thread blowing up WAY more than I anticipated (literally, yesterday, I was stunned to open my e-mail and find 30 some odd notifications), my intent was purely to see if others were running into the same things I was seeing, and get an accurate representation of if it was just me being whiny and not handling losses well, or if there really had been a skew like I had believed to be experiencing.

I thank you very much for taking the time to answer the inquiry I put forth to you as well :smiley:

1 Like

Right, I’m the one being obnoxious and sanctimonious. I’m not the one lying about what I said. We can keep going back and forth on this but your words are here for everyone to read even if you pretend to have no clue what you said. There’s no point in me continuing this conversation since you have no desire to acknowledge what you said and instead try to deflect, deflect, deflect. But by all means, keep trying to “stick it to me” if that’s what makes you feel better about yourself.

I have literally acknowledged what I said, even quoting it right back to you and explaining precisely what it means. You are the one refusing to read it and properly interpret it, instead choosing to only read what you believe was said and take that as truth and run with it.

And yes, you are being obnoxious and sanctimonious, you are LITERALLY the only person in this thread badgering me about what you “think” I said. Or have you not noticed that?

Indeed I shall if you continue to babble on in this thread, as long as it takes for it to eventually sink in that the problem here is your perception of what I said (despite literally quoting it back to me).

Way to gloss over how my saying I was sticking it to you didn’t prove any so called point of yours at all, however. Guess if you can’t argue it or give any kind of credible excuse, you just ignore it. Good to know!

1 Like

This is news to me, and quite interesting. Thanks for the reply. @Sirrian, when you say “decrease the chance of an extra turn”:

  1. Does that mean specifically preventing a 4+ match from dropping and matching automatically, or also preventing an immediate manual 4+ match from being possible with the dropped gems?
  2. Is the chance of an extra turn by default entirely random before 5 extra turns? (As in, you don’t mess with the air drops at all before then?)
  3. Is there a one-off threshold at 5 extra turns at which point the combo breaking turns on, or does the chance of combo breaking increase progressively after 5,6,7… extra turns?
  4. Do all sources of extra turns (including goblin spells, etc.) count toward this threshold, or is it specifically extra turns due to 4+ matches?

I’m glad the topic has yielded a discussion that provides some insight into the game mechanics.

4 Likes

Yeah it’s just hard to tell someone that thinks they’re seeing something with their own eyes, that they’re not. That’s what it comes down to. We just aren’t wired that way.

You just have to tell yourself my eyes are deceiving me or you don’t believe the Devs either way. It’s nice to see the Devs respond, but for some it will never help. Which is understandable when you look at it in that perspective.

1 Like

Actually, I take issue with those who think the Devs are misleading or lying to players; those that just think the Devs are mistaken I have no issue with. I don’t see it personally but that doesn’t mean the game isn’t messing up in ways I’m not noticing or maybe only affecting certain players.

1 Like

Here are my actual results against TDS today:
11 straight kills with no resurrections
1 resurrection
1 kill with no resurrection
4 resurrections in one battle
1 kill with no resurrection

I didn’t use my dragon team much today and when I did, they were easy wins, so no results for my own resurrection chances, but the AI’s results were, on average, about what was expected (5/18 = 28%). I acknowledge that this is a pretty small sample. Now that I think of it, my math is off. I didn’t count the expected death after a resurrection as an event, so it should actually be 5/21, which is less than 20%.

Even if I hadn’t been tracking TDS resurrections, that battle with 4 resurrections was crazy and would really stick out in my mind. The opponent had two TDS’s and at one point, I killed the first with Wulfgarok, which triggered a resurrection, then my Wulfgarok devour processed on the freshly resurrected TDS which resurrected again. Insanity.

However, if not for this conversation that we had yesterday, the 11 straight kills with no resurrections wouldn’t have been noticable to me at all. This is classic recall bias.

If you really think your results would be that different than mine, I’d encourage you to track your results for a while and see what happens. I’m pretty sure that if you get a big enough sample, you’ll find it comes pretty close to the stated percentage.

3 Likes

Well i had to kill them 11x in a game, that was more then 25% in my opinion

1 Like

It looks like you was frustrated from the way that camera was moving. :laughing:

110 life and 60 armor. Damn laughs

I think the highest I’ve ever gotten Jarl up to (my main) was 120 ish life. That’s some crazy high level playing going on.

It’s specifically targetted at only one thing… 4+ matches from gem spammers (like Rock Worm). No other types of 4+ matches are affected or modified in any way by this. Largely it’s there to stop AI looping teams becoming TOO frustrating… and I suspect it hardly ever even executes these days, now that we have capped all gem spam (except for Mr. Bone Dragon!)

3 Likes

AI combo break is nothing new to gems of war, and I recall you saying that the gem match / drop algorithm is basically the same from the original puzzle quest. So nothing has changed lately?

Yes - the AI is still basically identical to what we had 10 years ago in Puzzle Quest. If I had to list the differences, they would be:

  1. We have the previously mentioned Combo-Breaker (stopping gem spam giving too many turns in a row)
  2. We have the “AI Luck Limits” imposed in Quests, Challenges, Explore, & Casual PvP running Normal Difficulty
  3. Our gem-matching choices on CONSOLE ONLY now use a different system based on configurable color sliders for your PvP Defense
  4. The Gems of War spell-casting AI does not integrate quite as well with the gem-match-choice AI, so some troops struggle with that (e.g. Goblins sometimes pass up a spellcast, when they really should go for it, since they get an extra turn anyway)

Not since the console AI Update just before Christmas. The engineering team has been way too busy moving the servers & getting the next console update ready to tinker with the AI.

4 Likes