Batlad's "Post-hoc points"

keithburgun

Administrator
Staff member
#1
Idea #1 - High-Score - Historically most games - Rogue-likes, Civ, etc - have basically used a scoring system pretty much the same as that used in pinball. This is horrible and hopefully everyone's on the same page about that.

Idea #2 - Binary win/loss - I am very confident that the correct answer is to have a simple, binary win/loss condition at the end of a game.

Idea #3 - @Hopenager claims (as you've seen in other threads here) that we should use his "par" system (which to me is basically just a slight variant on Idea #1)

Idea #4 - @Batlad has this new idea for post-hoc points. What are those?


From the discord:

"Post-hoc points: These points are awarded on completion of an arc, based upon the remaining resources the player has left over. So as an example a upon completion of a match a player may get extra points based on having leftover time, troops, or other resources that were not used. Now keep in mind these bonus points are contingent upon the players success at the long arc. I think of these as rewards for efficient completion of an arc (efficiency points)."

"Post-hoc points:: a) are awarded on the completion of the longest arc in a match, b) measure some aspect of player skill or efficiency at that arc, and c) whose meaning is contingent upon the completion of the arc. The prime example of this is the Time Taken to complete a match. Speedrunning is a very popular way of playing games, and I believe, at least in part, this is because it represent a fairly good post-hoc point system. The meaning of Time Taken as a measure success is contingent upon the completion of the match. Before that, match is over the meaning of Time Taken is completely ambiguous. It could mean you're getting close to the end or it could mean you put the controller down and went to make coffee. Time Taken has no predictive power about whether you'll succeed at this match, or not. However, once the match is over, Time Taken suddenly tells us a bunch of extra information about how the match went. There were only a couple of seconds left on the timer, so the player just barely scraped through, this was one of your fastest ever times, you had minutes left, you seem to be getting consistently better. So the reason that post-hoc rewards are contingent upon the completion of the arc is that they have no predictive power about the success of an arc before the arc is complete, and instead they only measure quality or efficiency retrospectively."
Here is my response to this idea.

Overall, I hate to be harsh, but this sounds really bad, actually worse than Hopenager's idea and maybe even worse than traditional high score somehow? Here are my objections:

1. Speedrunning is popular, that's true, but you can't build your goal around speedrunning or even "efficiency" in a strategy game because then you flatten out the axes of strategy. I mean, does this mean that every match you should just rushdown? Shouldn't we want players to have room to make mistakes and suboptimal moves, so that they may also have the flexibility to make creative decisions?

2. Why are we rewarding the completion of a single arc with "points", at all? Doesn't the game already do that?

3. " However, once the match is over, Time Taken suddenly tells us a bunch of extra information about how the match went." - is that actually true? Couldn't there be many strategy games wherein the Time Taken to do a single arc could be both "longer than average" and also "better"? Shouldn't some strategies be slower?? Are we prioritizing rushdown over defense play for some reason?

What do you think?


EDIT: After some discussion with Batlad I decided that actually this system is not so bad, definitely better than Par and high score, but I am not sure it's better than just a normal win/loss.
 
Last edited:
#2
Thanks for taking the time to consider this idea.

First let me restate what I wrote on the discord to make it a little clearer.

Post-hoc points: a) are awarded on the completion of an arc, b) measure some aspect of player skill or efficiency at that arc, and c) whose meaning is contingent upon the completion of the arc.

The prime example of this is the Time Taken to complete a match. Speedrunning is a very popular way of playing games, and I believe, at least in part, this is because it represent a fairly good post-hoc point system.

The meaning of Time Taken as a measure success is contingent upon the completion of the match. Before that, match is over the meaning of Time Taken is completely ambiguous. It could mean you're getting close to the end or it could mean you put the controller down and went to make coffee. Time Taken has no predictive power about whether you'll succeed at this match, or not. However, once the match is over, Time Taken suddenly tells us a bunch of extra information about how the match went. There were only a couple of seconds left on the timer, so the player just barely scraped through, this was one of your fastest ever times, you had minutes left, you seem to be getting consistently better.

So the reason that post-hoc rewards are contingent upon the completion of the arc is that they have no predictive power about the success of an arc before the arc is complete, and instead they only measure quality or efficiency retrospectively.

Time Taken is generally a safe bet as a possible measure of skill and efficiency, but there are other possibilities depending on your particular game, such as resource usage, or challenges completed. So if you're deciding to make a game with post-hoc points, make sure that whatever you pick does actually represent player skill. And don't put it on a high score board, use par so that your game measures consistent skill improvements, not luck.


Now I think that for strategy game design, we're most interested in post-hoc points that only apply to the longest arc of a match.

If we're just talking about these match completion post-hoc points I think your last three question become irrelevant (but feel free to correct me).

Your first question is a lot more interesting though, it has two parts.

[We shouldn't measure Ed.] "efficiency" in a strategy game because then you flatten out the axes of strategy. I mean, does this mean that every match you should just rushdown?
I think this question suggests a few key misunderstandings of how strategies work and interact and how skill can be measured. To address this first let me say that I think the purpose of using post-hoc points, over binary win-loss is that they allow your game to more easily differentiate (and reward) players based on their skill.

Lets think about an example, say we have two players players that are at a similar level of skill, but player X has the extra skill of spotting opportunities to win early. Player Y is just as good generally but sometimes misses chances or is slow to switch to a rush strategy. Now I think that this makes player X a better player than player Y. A binary-win loss system will eventually be able to identify this difference, as long as there are some percentage of matches where player Y loses, when they would have won if they switched to rush sooner like player X always does when they see that game state. But, there are a large number of potential matches where failing to capitalise on this opening doesn't cause player Y to lose, it just makes their match take a bit more time than it otherwise would have.

What I propose instead is that we should implemented a post-hoc reward upon winning a match based on how fast the player won. The result of this would be that in cases where player X won with their rush strategy and player Y won with their econ strategy, we could give player X a few more points to contribute to their par. And as such, our goal system would incentivise efficiency by rewarding players that can push the system to its limits.

I hope this example illustrates why I think that carefully selected post-hoc point systems (tailored to the specifics of the game you're designing) can help your game to reward player skill.

I think another part of these misconceptions comes from the idea that your should always just use the fastest strategy that exists if it is rewarded, i.e. an aggressive rush strategy. But, this idea completely ignores the fact that you still need to win. Always rushing is a bad and easy to counter plan in good games and at high difficulties, it has low chance of winning overall. And if you don't win in a post-hoc point system you don't get any points, so winning is still the most important thing. Instead what post-hoc with a Time Taken does is to reward players who not only maintain a high win rate, but those that find opportunities to win sooner than what would normally be safe to try, by spotting opportunities in the game state that demonstrate the safety of an earlier than normal win in this match.

Shouldn't we want players to have room to make mistakes and suboptimal moves, so that they may also have the flexibility to make creative decisions?
I think this is a fair criticism of hard games in general. But I'm not sure it applies to post-hoc systems specifically. For example, binary win loss doesn't allow flexibility either, when those choices are sub optimal in terms of affecting your win percentage. In the same way post-hoc doesn't allow flexibility when those choices are sub-optimal in terms of their average effect on par. Really the only difference here is in extent, post-hoc pushes players to be as good as possible all the time, where as players won't notice this as strongly until they hit their limits (or the maximum theoretical limit on difficulty) in a binary win loss ranking system.

But what is the point of any goal system but to tell players what strategies are not good enough, while steadily ratcheting up the lower limit of what strategies the game will tolerate the player to use.
 
Last edited:
#3
...if you don't win in a post-hoc point system you don't get any points, so winning is still the most important thing.
I don't think this is a good idea. In the same way that a binary system is unable to distinguish between a player who just barely wins and one who wins very efficiently and decisively, your system would be unable to distinguish between a player who just barely loses and one who fails spectacularly. A full score system without a notion of "winning" or "losing", where the player gets only a score at the end of the game, does not have this problem.

However, you could still use speedrunning for a score system in a game without winning or losing. Just make it so there is no possibility of "losing", so the game always in the player accomplishing their goal and the only thing that changes is how long it takes them to do so. For instance you could make it so enemies don't ever kill the player, but merely slow them down somehow.
 
#4
It seems to me that there are two different goals a competitive player might have, and for one -- becoming consistently good at a game, while still having a somewhat "normal"/fun play experience -- a binary-score + ELO system is best. For Auro, Invisible Inc, or Slay the Spire (games with a lot of depth and/or randomness) I'd want to play something like that.
But there's also the speed-running-esque appeal of completely mastering a (usually more shallow and/or deterministic) game. It can be neat to do or even just watch the all-time best play-through of a game. And for that, a post-hoc score seems best. This might not be best for Keith's type of strategy game, but for something designed to be played through once (an adventure or puzzle) but that still has some depth, this could be cool. Like beating a sokobon-like game (Pipe Push paradise, Good Snowman) in the fewest number of moves, or a Zelda-like game while taking the least damage possible.
 
#5
(Those latter types of games may not have the lifestyle-games quality that strategy games have, but are nice in that they respect the average player's time, while also providing a goal for those who become obsessed with the game for whatever reason.)