Judging the quality of stats
One of the most frequent complaints I see on the forums/IRC is that a particular round has 'bad' stats'. To some extent such judgements will be based on personal taste, but it has lead me to wonder - is there some set of objective criteria by which we can say a set of stats are better/worse than another set? Personally, I think if we could find some kind of criteria like this it would help us to be a lot more certain about which rounds really had good stats and which didn't, and in the end lead to better stats for future rounds.
To get things started, I'd like to propose a few things which could be looked at. I will begin by saying that I am in general sceptical of the ability of people to judge a set of stats just by looking at them; it may be possible, but I suspect it's just too complex a problem to accurately judge in all but the most obvious of cases. I would be interested in any ideas people have about such a method. Moving onto specific criteria, I think there are two things a good set of stats should have; balance and variation. People often throw around the word balance without really defining what it means. In my mind, there are actually two separate things which need to be balanced - races and attack/defence. Race balance means that each race has a roughly equal chance of doing well. People often look at the distribution of races in the top100, and I think this is a fairly good measure, although you should probably look at the top10/50/200 as well, and perhaps normalise by the total number of each race. For attack/defence balance, this means roughly how easy it is to land/stop an attack. This can be a little harder to measure, but something that could be looked at is the number of launched attacks which land and the number of launched attacks which cap. If further data was available, you could look at for each withdrawn attack, what ratio of defending value to attacking value was necessary on average. For variance, there are again a few different things you could look at. I think you could break this down into per race ship strat and overall strategy. For each individual race, you'd ideally want a least 2 feasible ship strategies and maybe more. If data was available about the total distribution of ships in the universe, you could try to do some kind of clustering and then count the numbers. For an overall strategy, you could look at doing something similar within statistics like # of attacks launched, # of def fleets launched, # of FCs/MCs/Dists, Gov. choice, etc. Again, you could try a similar kind of analysis - perform clustering and count the number of clusters. While figuring this sort of stuff out isn't trivial, I dont think it would take a massive amount of work either; much of this can be calculated using standard software libraries. The main obstacle as I see it is lack of data; We already get some random stats about the universe, it'd be really nice if we could see this sort of information included for future rounds. |
Re: Judging the quality of stats
Quote:
|
Re: Judging the quality of stats
Quote:
...OK, fine. My point stands, but take into account that the remainder of this post is probably not objective holy truth but just my best attempt at such. Quote:
Quote:
Quote:
Quote:
I think you've identified the three main points: balance between races, correct relation between attack and defense, and a variety of ship strategies for each race. I'd like to see a wider range of possibilities on that last point, though this is not possible in the current framework for ship stats: What I would like to make possible is a true evolution of ship strategies as the round progresses. Right now, almost all ships are available on the first night out of protection, and all of them are on the second night. You can pick a ship strategy at tick 0 and stick with it for the remainder of the round. That's pretty static (that's a euphemism for 'boring'). We could delay the research on certain ship classes until much later in the round. Release FR/DE at tick ~400 and CR/BS at tick ~700, making each advance significantly stronger than the previous classes. You can either save up resources, crippling your early round growth, then get a growth spurt near the end of the round to make up for it, or spend on Fr/De, grow slowly and steadily throughout the round, or go for Fi/Co, sprint to the front of the pack early on, then work the rest of the round to maintain your lead. That's not the only way in which such evolution is possible. You could provide the ability to upgrade existing ships through research. I'd assign different limits to upgradability to each ship class. Make some classes strong with few upgrades, some weak with many upgrades, and some in between. Like the above, this allows you to choose a longer term strategy than is currently possible. This does lead to an explosion in the number of 'ships' (each base ship and each of its upgraded forms could be considered a ship in itself), and that the number of ships is already too high. Fortunately, there is no real reason why each race needs its own set of ships. We could combine them all into one big pool of, say, 18 combat ships (each with 1-3 upgrades), 6 pods and 2-3 SKs, instead of 40-50 combat ships, 10-15 pods and 5 SKs, split into 5 mutually exclusive sets. Quote:
|
Re: Judging the quality of stats
Quote:
|
Re: Judging the quality of stats
Quote:
However, any analysis like this in general will always be impacted upon by the politics of the universe - if Alliance X goes ship strat A and loses heavily to alliance Y with ship strat B, does that mean B is necessarily better than A? Probably not. If there is a surplus of BS fleets at the moment, is that because BS are overpowered, or because they are so underpowered, nobody built any anti-bs? Regardless of that, my feeling is that by averaging over the whole round you'll still get enough useful data out to improve the current situation. It won't solve the problem of designing a good set of stats completely, and you'll probably still need to apply some intuition, but it will be an improvement over the current situation where people mostly apply guesswork. |
Re: Judging the quality of stats
Quote:
I also understand that any final judgement about the stats will necessarily be subjective - even if we have a whole host of well defined quantitative measures, there will be personal preferences which determine how you go from say attack success rate to good/bad. Nonetheless, I still think such measures are useful. The issue you mentioned - why do multiple target rounds feel more defensive - is the sort of thing this kind of analysis may give us some insight into. The main way these measures give us this insight is by allowing proper comparisons between stats. While these comparisons wont be 100% accurate, they will be better than the current system of guesswork and public opinion (I think). It's interesting that you said the databases for prior rounds have been saved - any idea if/how the community could get access to them? |
Re: Judging the quality of stats
These stats are pretty balance EOD or are we still complaining Xans are OP?
|
Re: Judging the quality of stats
Quote:
To some extent, I would argue PA has something like this naturally. Rather than being driven by ship stats, it is driven the the popular opinion about the ship stats. This round for example, I started off with quite a lot of Fr and was unsure whether I would get many BS or not. Pretty soon it became evident that Fr was not particularly effective and I switched pretty hard into a BS attack fleet. Now the round has reached the midway point, BS are becoming much harder to land, so I'm moving into Co class. Nonetheless, I would agree that strengthening this aspect is probably a good thing; you would of course have to be careful to manage the complexity of such a system, and you'd also need to be careful not to make the early/mid game too limited for late game players. Not many people will sit through 850 ticks of being useless in my opinion. I'm not sure I can get behind the specific plan you mention though. There's nothing wrong with it in principle, but I'm not sure the will/ability to make the required changes to the code exist. I think an easier to implement change which may lead to a similar effect is to expand both the tech tree and range of construction options. At the moment, most planets will finish the majority of the tech tree, and certainly finish all the 'must-have' techs. If the tech tree was expanded such that no planet could finish more than 70% of it, then people would have to make strategic choices about where to invest their research. I haven't given much thought to what such a tech tree would look like, but you could for example extend the hull research branch beyond cruisers. If we allow branching, we could have research that improves the armor/damage/emp-efficiency of all ships. If each of these branches had 2-3 levels and each required say 14000 RP, people have serious choices to make and the landscape of the game would change as the universe finishes each branch at roughly the same time. |
Re: Judging the quality of stats
Quote:
|
Re: Judging the quality of stats
Quote:
|
Re: Judging the quality of stats
Quote:
Quote:
Quote:
Quote:
Because you're right; unlike in StarCraft games, PA rounds take ages. You may want to be willing to wait for Ultralisk tech to finish before destroying your opponent, but no sane person would twiddle their thumbs for weeks before something interesting happens. Making the upgrades fairly low key ensures that even choosing a fleet that's relatively weak to start with does not immediately disqualify you from playing the first half of the round, and that choosing a fleet that has a relatively shallow pool of upgrades does not stop you from enjoying the second half of the round. In any case, those are just the two extremes. Maybe fleets could upgrade in distinct steps: Cr gets its upgrades around tick 50, 300 and 700, while Fr gets its upgrades around tick 100, 400 and 600. Or maybe if you went Cat, you could get your Cr upgrades at tick 650 instead of 700, preventing Fr from smacking you in the face for too long. The possibilities are endless. Quote:
Quote:
I don't think your approach is much simpler to implement than mine, though. You still have to add the research for a bunch of upgrades, some means of presenting the different versions of the ships in some sane way, the combat engine should take them into account... all of those things are hardcoded, so it's far from a small addition. Unfortunately, both of our ideas venture so far outside of the realm of practical possibilities (given the lack of developer time) that it's nearly pointless even taking cost into account. In any case, while this is all fascinating, none of it has anything to do with your original question. Sorry for derailing you. :P |
Re: Judging the quality of stats
Quote:
Back on topic; without trying to re-invent the "wheel" that is PA, I don't see any issues with reusing previous stats (with some tweaks) if it makes sense to do so. Constantly starting from scratch each round has proved to be problematic. Look at other games, and how their "updates" are handled...it certainly doesn't involve such drastic changes to things in-between (insert "patches", "updates" here). Might be apples and oranges, but I think the point remains. PA has painted itself into a corner, and there's no easy way out. Keep tweaking stats each round. Shift from MT to ST stats. Limp slowly on. |
Re: Judging the quality of stats
Quote:
Quote:
Quote:
To judge stats based on this will be a good round or bad round is impossible as they only play a small part in the round. However, when the stats are highly defensive and you can clearly see you can hold roids, they are the worst set of stats impossible as it limits solo play, makes blocking stronger and lets the better players just sail off into the distance. Quote:
Quote:
Is Fi/co strong? if Yes/Average, go Xan Fi/Co with whatever other races is ok, usually cath. Quote:
Heres the problem as well, say we work out and get a definitive list on round stats: R59: Good R58: Bad R57: Best stats ever R56: ok and we use this data and select R57 as the next stats to use, It makes very little effect on wither the round will be awesome or bad as alliance politics will come into play and the round will end up different. |
Re: Judging the quality of stats
Quote:
Everyone starts with Fi ships unlocked (Why force research onto something everyone needs), then as the round goes on, each set of ships is slightly stronger than the previous. However, each race has certain ships which pwn in each category. Traveltime for everything is set at CR/BS pace, no more different travel times and allows for proper shipstats where you can try anything. Some races may only need to go De, while some might want BS as its awesome and wtfpwns everything. Problem is, PA has no coders so on it ticks with 0 changes. |
Re: Judging the quality of stats
Quote:
(To be fair, you did go on to expand on that in your later posts). |
Re: Judging the quality of stats
Quote:
I think that the point of this thread is redundant because it's trying to solve a problem that doesn't exsist, just one that people who screwed up their choices think does. Defensive/offensive is just a strategy and play style choice it isn't hardcoded into the game. Alliances decide how attacking and defence minded a round is..... |
Re: Judging the quality of stats
Quote:
Quote:
* If the ship stats only had pods in a single class, then they'd encourage offensive strategies, because all forms of defense are objectively impossible, even red defense: no one would prelaunch (so scans are useless), and no one would need to tell others about when and where they're attacking (so spies are useless). * If, besides those pods, every race also had a 0 loss defense ship that fired at the one roiding class, then they encourage defensive strategies, because every landing will hurt the attacker. This demonstrates that in their most extreme forms, the stats can make certain strategies flat out impossible. In milder forms, they merely make certain strategies less profitable than others. Stats influence how the round is played because players and alliances will naturally gravitate towards the strongest strategies (even if those are not necessarily the most entertaining or the best for the long term viability of the game). The function of the stats is to encourage forms of play that are both fun and long term sustainable. Stats that encourage overly defensive strategies are boring because they turn PA into an initing game. Stats that encourage overly offensive strategies are bad because they corrode the bonds of community. Quote:
Since you're not actually arguing against Haer's points. Do you agree or disagree that the stats are better if the races are roughly equally strong? Do you agree or disagree that the stats are better if they're not overly offensive or overly defensive? Do you agree or disagree that the stats are better if each race has more than 1 ship strategy available? If you're not willing to answer any of those points, you're just polluting this thread, and I would ask you to kindly get out. |
Re: Judging the quality of stats
Quote:
This is the point I was trying to make about the post being redundant. If you were to ask me how to 'balance' the stats, and by balance I mean make equal amounts of players play each race then the primary solution has to be the removal of ETD. Having to correct the targetting to give an even spread with 5 races has proved a big stepping stone to a lot of stat makers and is normally why we end up with 1 race being over picked compared to the rest. As a final point tho having a set of stats where 50% of the playerbase goes 1 race and 1 class in that race doesn't make those stats bad. It just makes that round weighted towards that races play style. As long as that race is weak or strong against itself then everyone is in the same boat and that is OK. |
Re: Judging the quality of stats
Quote:
|
Re: Judging the quality of stats
Quote:
|
Re: Judging the quality of stats
Quote:
|
Re: Judging the quality of stats
Quote:
I can go on, but i think this might lit a bulb or two. |
Re: Judging the quality of stats
Quote:
|
Re: Judging the quality of stats
Quote:
I am interested in the judging the stats from the perspective of the second question. Your proposed method works for the first point of view, but doesnt tell us much about the second. |
Re: Judging the quality of stats
Quote:
But knowing the game in an abstract sense should be enough to create a set of stats that has a good chance of influencing the round in a positive way. If you don't think that's possible, then why shouldn't we just generate the stats completely randomly? |
Re: Judging the quality of stats
Quote:
You talk of cause and effect and reckon that stats influence the rest rather than the rest influencing stats (or atleast how the stat set plays out), that is completely false. The whole matter of an alliance picking its strategy based on its political intel of what other alliances are doing and basing its race choices and class choices based on what others are doing means that stats are always open to political influence from even before tick 1. There has been rounds in the past where a race considered OP has faltered because it did not fit into the political plans of alliances in that current round. Who is to say if that set of stats was re run 5 rounds later that the OP race wouldnt be picked by 50% of the playerbase as the layout of the universe changed and drastically change how those stats play out as the distribution is different. MCs have changed the playstye of a lot of players and some alliances drastically in the last 3 rounds let alone the last 10 with all the other small changes PA Team has introduced (look at how underplayed Zik is now salvage has been smashed around repeatedly). I actually like the idea of RNG stats. I think it would be a breath of fresh air to the game as their would be no influencing of the stat maker by a select few and no fiddling of stats which end up changing a creators original idea into some horrid mess. I would actually go one step further for balance and cap the amount of each race that can be picked by the playerbase. 20% of each, there is 2 weeks of pure downtime inbetween rounds so there is no excuses for missing signups. Intial cap at 100 (low end expected turnout of players) rising by 20 after 400 planets have signed and again every 100 etc. Taking all the human influence away from the stats would finally remove the flaming/trolling/whining i believe. Blame the machine, the machine doesnt care.... |
Re: Judging the quality of stats
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
|
Re: Judging the quality of stats
Quote:
|
Re: Judging the quality of stats
Quote:
Otherwise I agree with your points, and am especially interested in not having everything available with the first 3-4 days. |
All times are GMT +1. The time now is 20:20. |
Powered by vBulletin® Version 3.8.1
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
Copyright ©2002 - 2018