Helping Survival Mode... to Survive... -
First, let me say, I think Survival Mode is a great mode to give value to old cards in perpetuity.
But, there's a problem: when the strongest players can't find opponents and players are intentionally losing rating to earn more? That tells me something isn't right.
So we are back to solutioning....
Liquidity Bots? Yeah, we tried that...
I was thinking about the liquidity bots in survival…
I mean, there is really no way to tune these bots:
- a) if you make them pretty weak, rating inflation gets way out of hand
- b) if you make them pretty strong, most players will be too weak and avoid them entirely
- this means that survival mode mostly becomes strong players vs weak bots, so go back to A).
- c) if you make them too hard, then the top players will start burning all their best cards against bots with unlimited decks, and players with strong decks will be no better off than players with weaker decks, which doesn't really seem right.
At the end of the day, bots with unlimited cards essentially defeats the entire concept of the mode - which is to outlast your opponents.
How can you outlast a bot that never loses cards?
Suggestion: Cut the bots, but tweak the matchmaking...
So let's operate from the perspective that bots with unlimited cards are antithetical to the concept of survival.
Let's say we kill the liquidity bot program in that mode.
But we still want to make sure that players can find matches, because that was a problem.
So let's look at the matchmaking algorithm.
Right now the algorithm tries to match players with a suitable opponent. If no one near your rating is found, then it waits.
The Survival mode matchmaking algorithm could do a simple match of the Waiting Queue from top rated, down.
So at any given point, the top two players in the "waiting queue" get matched. Doesn't matter if they've played each other recently - just match the top two, then the next two, then the next, and so on, in batches.
How does this help?
- The top rated players will always find a match against the hardest player available at the time
- Meanwhile, most players in the middle and bottom will usually be matched against player with a pretty similar rating,
- There are no bots with phantom decks that are laughably easy or horribly overpowered,
- Someone's cards are getting disabled every match, and
- no long periods waiting for match, unless there is literally only a single person in the queue
You are now thinking:
"Oh No! Some whale is going to play at the same time as me, and disable all my best cards!"
Carry on, Constant Reader: I do have a suggestion to address that.
Avoiding Repeat Matchups
Ideally, we want to avoid players having repeated rematches. If the top two players keep queuing up at the same time, that might happen. (Especially if they are both using a bot provider which doesn't spend time watching matches).
When I looked at last season's leaderboard, I saw the majority of survival leaderboard players had a few hundred matches in the season, so 10-30 matches per day.
The busiest players had a few thousand matches all season, averaging a few hundred matches per day:
- instamental: 4423 battles
- steem-eng: 2505 battles
- bamlolx: 2499 battles
- dathomir: 2373 battles
- nichtwiesie: 2368 battles
- vugtis: 2278 battles
- waterisall: 1855 battles
- tryhard65: 1854 battles
- dratek: 1549 battles
- bluewolfz: 1492 battles
- captainwhoco: 1356 battles
- sweenyswift: 1340 battles
- milomex: 1326 battles
- bravetofu: 1307 battles
- bltzr-wizard-701: 1284 battles
- neal.cards: 1282 battles
- twisted-jester: 1143 battles
- dejota: 1127 battles
- michaelb: 1091 battles
- cjturtleman: 1023 battles
So the busiest player averaged around 13 matches per hour.
Here's the idea:
Survival could implement a short, randomly determined battle cooldown for each opponent after a survival matchup.
(I'm thinking something ranging from say 30-240 seconds after matchup)
Most manual players won't notice this, because it takes a couple minutes to select a team, choose tactics, and see the results and laugh or cry maniacally.
So no pain for humans.
Meanwhile, bots just queue, wait, submit quickly, and requeue: they won't really be affected by a short wait either.
So how does a randomized short cooldown help?
Let's say the top two players match up, but Player A has more dominant cards and wins.
So Player A gets a random wait time of 153 seconds, after which they can rejoin the matching queue.
And Player B gets a random wait time of 103 seconds, at which point they rejoin the matching queue and immediately are matched with whoever is at the top of the queue before their last opponent is able to rejoin the queue. This should help avoid repeated rematches.
Last Piece: Control Rating Inflation by bringing Elo rating back to Survival
Last thing - we used to have an Elo rating in ranked. We simplified it to +-20 a while ago.
But this proposed system might result in more frequent top players facing much lower ranked players in periods of low liquidity.
I don't think it's unfair, in a mode called "Survival," that the biggest, deepest deck stands on the top of the mountain and slashes whoever comes closest. But I don't think that they should get as much rating from beating up the little guy as an equally powered opponent.
And I think if the underdog wins, we should reward them with better than a flat +20 rating, possibly even a boosted reward share - (suggestion: could perhaps use rewards halfway between the ratings).
Anyways, this is just my entry in the "let's get the community to give us unsolicited ideas to fix things" category!
Hope you have a Vonak day!
#WenBeaver
Best idea i have seen so far! - thanks for sharing
Thanks Bam! I think Survival is a great mode; I just wanted to try and help with (hopefully) realistic approaches to keep it going
Other games do use the same strat for high rating players...its a proofen strategy, in my eyes the best way.
I like to test leagues, bronze to diamond, like it was before in ranked in survival. They will have different cooldowns based on leagues. Lower the higher number of days. People can choose which leagues they want to be. Can can be matched with higher or lower league 25% of the time but 75% of the time they are matched within their leagues.
And I also like to try your option fir 2 weeks.
Yeah thanks, imagine we had enough resources to running a couple different modes side by side for a season or two - that would be some serious A/B testing!
I'd also be really interested to see modified versions of leagues, and different cooldowns seems like a good approach.
I can live with one after the another test, tagging @royaleagle to read this.
Thanks for sharing! - @azircon

Congratulations @oaaguy! You have completed the following achievement on the Hive blockchain And have been rewarded with New badge(s)
Your next target is to reach 1500 upvotes.
You can view your badges on your board and compare yourself to others in the Ranking
If you no longer want to receive notifications, reply to this comment with the word
STOP
good thoughts @oaaguy, but this leads to other problems. I'm happy to have a discussion with you about it.
I do agree with many parts of your analysis, but there are other parts you are missing.
No worries Dave, I don't have all the data you do. All I wanted was to help put some ideas on the table. As it's been considered and either rejected or refined, that's just the process at work - thanks for the reply!