Player Features and Ideas Discussion

OK so CCP occasionally rebalance, nerfing stuff and boosting stuff. Sometimes in doing so they screw up, which leads to other races/ships being unbalanced,a nd more rebalancing needed. Can that situation be improved upon?

Q. How do you know a ship is unbalanced?
A No one is flying it, or everyone is flying it.

Q Why is noone using it?
A It has no useful role, compared to the other options

Q Do you have to get a rebalance completely correct?
A No, its fine to close the gap. Its less fine to over compensate.

CCP have at their disposal all the stats for ships built/ships lost, when why and how. So it should be possible to rebalance ships in an automated way - easily between races, and with some thought between ship classes too. Thats this suggestion. I don't want anyone to focus on the specific example below, but think of the overall approach.

Example recipe:

Convert all ship bonusses to continually varying attributes (e.g. things like 25m^3 of drone capacity per level, need to be converted to drone HP or drone damage etc - or ignored)

Sort all ships into a class (we aim to balance between ships within that class)

Decide what you *want* the distribution of ships in each class to be (presumably, you want races to be balanced as the minimum, but you might want to, say, balance between destroyers and cruisers too, if you think destroyers as a whole have no use)

Choose a metric for deciding current ship popularity. Prefferably something that isn't skewed by high sec mission runners etc. Say ships lost in 0.0-0.4 systems, or ships constructed, or sampled actual usage.

Decide a maximum change per iteration, this contains the impact of changes.

Within that class make weighted adjustment to the ship bonus values, from their current values.

Repeat every 2 months.

Example:

I want all cruisers to be usage balanced across races, and I want 20% of cruisers to be T2. So my target (where G1= Gallante Tech 1 etc) ) looks like;

G1 A1 C1 M1 G2 A2 C2 M2
0.2 0.2 0.2 0.2 0.05 0.05 0.05 0.05

I look at the usage profile, and see popularity distribution like;

G1 A1 C1 M1 G2 A2 C2 M2
0.03 0.1 0.4 0.1 0.1 0.1 0.07 0.1 (hopefully adds up to 1.0)

Looks like there are too many T2 cruisers, and galante T1 is crap, but Caldari T1 is way overpowered. I now look at the discrepancies.

G1 A1 C1 M1 G2 A2 C2 M2
0.17 0.1 -0.2 0.1 -.05 -0.05 -0.02 -0.05

and scale these to the maximum allowed boost say 5% (ie mutliply by 5/0.17=27.41 in this case)

Choose a bonus randomly from each ship, and apply this correction to that bonus.

Here out G1 figure is 5% (worst offender) and C2 is -0.54% reflecting the G1 ships are massively underused, and C2 ships are only slightly overused.

So picking two example ships, we make the following change (from the current values):

thorax: Gallente Cruiser Skill Bonus: 5% bonus to Medium Hybrid Turret damage per level and 5.25% increase to MicroWarpdrive capacitor bonus per level.

cerberus: Caldari Cruiser Skill Bonus: 5% bonus to Kinetic Missile damage and 10% bonus to Missile velocity per level, Heavy Assault Ship Skill Bonus: 9.94% bonus to Heavy Assault and Heavy Missile flight time and 5% bonus to Missile Launcher rate of fire per level

obviously this is just based on made up numbers, before any cerberus owners get indignant! and so on for each ship in that class. And for each class.

Pros:

Automated rebalancing decreases load on CCP manual effort to rebalance

Small regular changes allow evolution and plenty of time to intervene if something becomes stupidly overpowered at some particular function

No ship/race stays king of the hilll for long - if its too popular, it grows less powerful

No need for module rebalancing. Ships get better/worse until the become useful

Can use to steer relative popularity of ship classes (e.g. set a target for cruiser vs. BC ratios)

Random element causes races to diverge a bit.

Cons:

Need to think about the bonuses and the ones that aren't continuously varying.

Somewhat lacking in backstory explanation

Usage will lag reality by virtue of training time - more significant for larger ship classes, and possibly BS+ should happen on a 4-monthly cycle.

Classes need to be carefully defined by role (ie comparing an AF with covops frig in the same class makes little sense, and could lead to stupid overpowering)

There will be winners and losers

In terms of correcting a specific deficiency, say 'hybrids are pants', different ships would compensate in different ways, some via RoF, some via range, some via DPS etc. Ships that don't have a bonus that actually helps, would develop different strengths via their *other* bonuses to compensate.

tl;dr. Just tweak the ship bonuses every two months according to how much use they are getting, using an algorithm that doesn't cause overall capability inflation or deflation. No matter how crap a ship design is, given enough time, it will be boosted enough to be competitive (and vice versa).

you ******* idiot this is stupid?


no, the ships shouldn't all be balanced. That would get BORING really (and I mean REALLY) fast. There just needs to be a better reason to have the smaller classes (or less-used hulls) around in a fight...

The motivation for suggesting the bonus that gets tweaked is randomly chosen (and for using the bonuses at all as the vehicle for balancing), is to ensure diversity. Ships should not all evolve into the same replicas of each other, but they should evolve into some rough balance of popularity, even though their roles and uses are much different.

Going back to the example. hybrids suck, therefore its likely under this that gal boats get boosted. Each ship has reasonably unique bonuses, so for example the Myrmi may develop its drone strength to compensate (which is unique) where the Brutix may become a (better) glass cannon. All the algorithm does is strive for popularity balance, it doesn't strive for role balance.

If one ship becomes powerful, but remains unpopular, then those who can figure out how to make use of it, will get a short term advantage (good on them!) whilst the masses play catch up. It should encourage a generally shifting style of play, without the game stagnating into 'all about short range', 'all about speed', or 'all about sniping' strategies. The ground will shift slowly over the course of 6-12 months.

What it most certainly doesn't do is modify hybrids so that they are similar to other weapons systems. IMO, this approach DIVERSIFIES play styles without favouritism or CCP deciding what constitutes 'doing it right'....

Boring might not be the right word. It's more like "confusing as hell" and "exploitable to insane levels".

First off, there's the fact that every single person who ever tried to fit out a ship or learn about ship's capabilities will hate you for all time. It's hard enough to learn the ships and fit them well as is, much less if you have to re-memorize all of them every month.

Second, this system would be extremely vulnerable to exploitation, no matter how many safeguards were put in place.

Third, there aren't really many ships that are actually OP in Eve (supers in the current state of Eve are a bit of an exception due to their massive DPS, EHP, and the logoff timers). There are ships that are better against other FotM ships, but that doesn't make it OP. Take the Dram, it will win when well-flown against many other frigs, but if it tries to fight a Daredevil or Cruor, it's in trouble. There's a RPS mechanic in Eve, there are very few ships that can't be taken down if you have the right ship yourself.

Instead of complaining about certain FotM ships being OP, go and try to figure out a counter to them and then beat all the FotM people. Once more people catch on to the kinds of ships you're flying, that will become the new FotM and the cycle will repeat itself. It's how Eve works, that's the beauty of the sandbox game.

Also, your proposition would completely remove this sandbox facet of Eve. Due to the constantly changing ships, rather than going out and trying to figure out how to beat the FotM, people would just wait it out and then try to get on the leading edge of the next FotM from the next rebalance. This would create MORE FotM, not less, since the constantly changing ships would devalue the effort of making a unique and effective fit.

Confusing as hell: You have to stay on the ball - but 2 months isn't set in stone. However CCP's normal timescale is too long!

Exploitable: Extremely interestind in hearing about specific exploits so that I can adapt the proposal

I can think of a few, for example the risk of something growing a mission-indestructible tank - however in theory this would be at the expense of dps, and would also fix itself with time.



haha. So there are two things here. The frequency... perhaps 2 months is two often? The amount of the change (my 5% scale above is a 5% change to the bonus amount - ie the max change ito a 10% bonus is to a 10.5% bonus, and most ships would see nowhere near that)




Overpower and underpowered are the same thing. My approach doesn't change the the overall power of the class, where an approach like CCP take generally leads to capability inflation (which is bad). Or basically if only one ship is really bad, the nerf to the others is very modest. If ships are generally equal, then nothing much changes anyway.



Whilst its true, its a statistics game. The algo takes care of this, it will sort out whether a ship is genuinely overpowered or not.



That is a fine approach if things are genuinely balanced and its purely a FotM issue. However there are some failships, some ships never make FotM and never will without balancing. Yours is an argument against balancing - mine is a strategy to do better balancing once you recognise the need for balancing in the first place - which I think many do - like there aren't hybrid threads everywhere .

Really?

Well what I consider boring month after month is this: http://www.eve-kill.net/?a=top20

-Where is the interesting stuff?
-Where are the options?


Even pods are more often in those stats than majority of ships in the game

So you'd prefer them to waste time writing code for what would need to be an extremely complex and intelligent AI system to do it for them, and then for said AI system to likely cause significant balance issues when it "balances" things?

This is exactly my point though: it won't work. Its too simplified. The only way for this to even come close to potentially working is to have a complex algorithm that takes into account a ton of factors, not just the values you have listed in a statistical approach.



Don't be stupid. You've put forward a mildly decent idea, no reason to ruin it with a terrible ad hominem attack based on the assumption that the algorithm I was referring to would only do those simple calculations.

For this to work, a much more complex algorithm would need to be devised; ship balancing is not so simple as the stats you have put forth here.

The core problem is utility vs. damage and tank everyone see hurricanes and drakes this day and age, due to the fact that the other battle cruisers lack the utility or the tank of these, if you look at assault ships what do you see flown the most, the Jag or the Ishkur the Utility and damage these ships is disproportional to that of the others, CCP desperately needs to revaluate the track, cap usage on turrets, tank and ewar of all there ships. The solution is not nerfs or bonus to the damage of ships but an increase in utility on ships that are less commonly flown I would like to see battle clinic compile a statistic on the class of ships flown in kills and loses so that we can see the uses of ships. The clear fact in any game is that people will use the flavor of the month because they want to be successful, when a ships like the hurricane is being flown in 75% of all kill we need to look at the other ships and bring them to par utility wise do not nerf these ships that will just lead to more Imbalance bring the other ships up with utility, a key example of this is the command ship they are flown less and less, when a hurricane can beat an Astarte there is a problem.

I'm a professional programmer - and I've worked on far larger and more complex systems than Eve. How many lines of code do you think it would take to implement the algorithm I suggested, including the db pulls and pushes? Remember this isn't an integrated component of Eve, its a standalone app that updates the db during a downtime.

Evaluating the complex change impact is hard in a system with as many degrees of freedom, maybe impossible - thats why you do it slowly, in tiny incremental changes, so you can throttle back/roll back and take action at the first sign of problems. In contrast. CCP will take a massive swing at hybrids this winter, and they also have no way of seeing the effect (rather like they didn't see the effect of the web nerf that created the problem!)