Making games more unpredictable

[cashto]

QUOTE (badp @ Dec 21 2010, 12:05 AM) As we all know, SgtBaker has discovered he can predict the outcome of 85% of the games by merely looking at who's playing against who -- without even factoring who's commanding.

I don't know how much I believe this factoid. If true, it tells us more about how stacked the average game is, and less about how accurate AllegSkill is.

If yellow has 100 AS and blue has 101 AS then Allegskill "predicts" that blue will win -- but not by a huge margin. Maybe there is 52% chance blue will win and 48% that yellow will win. It's pretty even.

So if AS accurately predicts 85% of games, what does that mean? Either 85% games are so stacked that Allegskill can predict the outcome with 100% certainty, or that Allegskill is so psychic that it can predict the outcome of every game, even the 100-vs-101 ones, with an accuracy of 85%. Or some combination of the two.

Of course Allegskill doesn't need to factor in who's commanding. The stack knows who the comms are. Just look where the stack goes tells you everything.

It may not be so bad, though. If the game is even for 45 minutes, then yellow blows blue's techbase, all of blue's vets ragequit, and yellow's newbies take 15 minutes to resign, Allegskill will count this as a "stacked" game ... even if it was even for all the time that mattered. Allegskill isn't so much "predicting" who will win given a team's current composition, as much as telling you who won based on the quitstack at the end of the game.

Wish we had much more transparency into the basic supposition of this thread before going off half-cocked on ways to "fix" it.

QUOTE Can Allegiance be made more unpredictable, maybe more unfair but also more interesting to play?[/quote]

I don't see how. More undpredictable = more random = making games less dependent on skill. If I wanted to play a game where skill doesn't matter I'd play Farmville.

QUOTE (notjarvis @ Dec 21 2010, 12:50 AM) To be honest I (personally) think shorter (at least slightly shorter) games in general might be good for Allegiance. One of the frequent reasons many people don't play is they don't unless they have a couple of hours free.

I don't agree with this. I think the pace of Alleg games is pretty reasonable, all things considering. Oftentimes the game depends on a steady performance, not a single knockout blow. Imagine playing a game where you really only get one or two opportunities to rush miners and the result of the game depends on who draws first blood. Killing a single miner shouldn't be tantamount to losing your opening con.

(Anyways the speed of the game has much more to with the size of the map, drone build times, drone travel velocity and mining rate -- and less to do with faction research times.)

What kills late night games is a succession of 20 minute crappy games, each one punctuated with an interminable delay trying to find even comms and pressing AFKers into joining one side or the other. Sure, three-hour snoozefests are just as bad for player retention, but they are much more rare, and at least there was a game that, for at least most of the time, each side thought they had a chance of winning if only they would get their act together.

[lexaal]

QUOTE (cashto @ Dec 24 2010, 10:22 PM) ...

Cashto'S post has 100% of that in it what i wanted to say.

[Spunkmeyer]

Yeah pretty much what cashto said.

Also AS is not the end-all-be-all skill indicator... it's very easy in Alleg to repeatedly end up in lost games or won games regardless of your skill level (especially if you are the type that doesn't care a whole lot about your rank, or if you always join big games where your contribution means little), and if that happens at the beginning of your "career" then it has far more weight on you ranking than later results. Conversely, you can work the system by making sure the other team has a weaker commander or a number of players who are overrated by AS. Or just make sure your faction is superior with money settings stacked against your favor. I'm sure it works great in 8vs8 DMs but Alleg is a bit more complicated than that.

[Makida]

I also agree with cashto, especially

QUOTE (cashto @ Dec 24 2010, 04:22 PM) More undpredictable = more random = making games less dependent on skill

and

QUOTE (cashto @ Dec 24 2010, 04:22 PM) I think the pace of Alleg games is pretty reasonable, all things considered.

[Camaro]

Shorter games already exist with RPS, you keep your economy functioning decently and your Assault level technology (with mk2 level guns) make short work of the other team.

[DonKarnage]

QUOTE (Camaro @ Dec 24 2010, 11:04 PM) Shorter games already exist with RPS, you keep your economy functioning decently and your Assault level technology (with mk2 level guns) make short work of the other team.

Now if we could only get people to play it with non-fufi settings.

[NightRychune]

RPS is probably the most balanced core ever for actual gameplay

[FreeBeer]

QUOTE (cashto @ Dec 24 2010, 05:22 PM) It may not be so bad, though. If the game is even for 45 minutes, then yellow blows blue's techbase, all of blue's vets ragequit, and yellow's newbies take 15 minutes to resign, Allegskill will count this as a "stacked" game ... even if it was even for all the time that mattered.

If I'm reading that right, and I've interpreted SgtBaker's notes on the subject correctly, I don't believe the comment is correct. AS calculates the team rankings on a continual basis, and is aware when people jump into the game and when they exit and pro-rates their skill value across the whole game duration. ie a 60 minute game that has a team AS of 200 for 50 minutes and 100 for 10 would have a rank of 200 * 5/6 + 100 * 1/6.

I recall SgtBaker and MrC discussing how to deal with late joiners and early exiters and how best to deal with it algorithmically. (or in MrC's case, Al Gore rhythmically)

[Spunkmeyer]

That is correct, that's why I bail as soon as a game starts going south

[FreeBeer]

That doesn't help, if I've read SgtBaker correctly.