Re-Organizing DSB

[falconeer]

East: 21.13904138
North: 20.16467189
South: 18.44031889
West: 16.77799676

Kills don't matter. These are years and years of practice stats? League was different. Many star players west have MVP and most kills and bell flags.

You should just use an average of the kills in each sector. The average is 19.1305. Though I'm unsure how kills matter. It should be based on amount of players in a sector no?

[jim the chin]

Well, you are right about one thing - I am lumping bonus points in with kill points, and then forcing that sum to be a ratio of flag points. But it's totally equivalent to lumping bonus points in with flag points, and then forcing kill points to be a ratio of that sum. It's also equivalent to creating a whole new category of "bonus points," and then forcing kill points to be some ratio of flag points and also forcing bonus points to be some other (possibly equivalent) ratio of flag points (which, as I understand it, is what you proposed when you said to treat it as its own category). They are 3 different, but totally equivalent, ways of looking at the same thing. The equivalence is captured in the quantities p and B.

If that's the case, then by including the kill bonus we're also reducing the bell flag points by an identical fraction.

With what you just described, you seem to be doing: "kills + kill bonus = flags" and suggesting this is the same as: "kills = flags + kill bonus" when it's actually: "kills = flags - kill bonus".

If the kill bonus is a completely independent quantity that doesn't encourage killing any more than it encourages flagging, then it doesn't need to be in this calculation. You would just pick a value for it that gives it the right importance.

Perhaps it's just easier to explain. I'd like to see something like:
Kills = 40% of all the points
Bell flags = 40% of all the points
Kill bonus = 20% of all the points (closest we can get to this number with B as an integer.)

Or we could do kills = 42.5%, bell flags = 42.5%, and kill bonus = 15%. That's what I mean by making the kill bonus an independent quantity - it doesn't affect the ratio between kills and bell flags.

It looks like you're holding bell flags at 50% before reducing kills from 50% to accommodate the kill bonus.

Is the problem that you're holding p=1? We want bell flags to be equal to kills (1:1 ratio), but not equal to kills + kill bonus, because the kill bonus is both a killing and a flagging quantity.

[Tembest]

Without looking much into the calculations, I noticed something wrong.

Entropy, the point is NOT to distribute bell flags and kills 50 % in a prac. That's impossible. The number of kills vary a lot depending on players, teams, etc. etc.
and it even depends on the scoring system itself (how smart it is to suicide, etc.)

So, don't look at average kills in a prac. It's completely irrelevant.

It is only about not overvaluing one over another at a ding moment. In other words, the only reason we look at the value is to give specials multiple purposes.

Like leagues 1-10 if you were about to die 40 seconds prior to ding with full specials, it made sense to. Then respawn again with full specials and use those specials for
flags. That's something we want to avoid, without over-emphasising kills either. So it's not about the total proportion of bell flags, kills or anything else, but that the game
could be played in multiple ways. 6vs6 league was a perfect example of this. First time ever using a repeller for a kill actually made sense.

[Tembest]

It's all about how many flags are worth dying for? It's not such a big deal.

We can set it at 2.5 or 3.0 for starters, and change it one way after 3 months...

No calculations we make will ever give us a perfect number. Gameplay changes the second we change those numbers, and we cannot know how.

That's why I'm proposing to use what we had in 6 vs 6 because it actually worked. Then we can start changing it to any direction we want.

As 'hello' perfectly stated, it's supposed to be a flagging game, and even if flags were worth 0 points, killing would still be important.
But of course the purpose is to find a ratio where it wouldn't be all about suiciding for specials. I remember the leagues I played south, south is so easy to break into that
after every ding, even if we had killed the enemy out, I attached to north or anywhere else just to get myself killed in hope of new specials. That's something the game
should not really be about.

[jim the chin]

Temb, I agree that's important, but what Ent's doing is relevant too. I suppose the easiest way to say it is: "the macroscopic reflects the microscopic". In other words the total ratio of kills to flags reflects all the minor occurrences that happen during a match.

However, I actually see your point, and agree with you that the scoring system itself may then change how people play. I share your desire for an `iterative improvement' to the scoring system so that if people change their strategy, we'll alter the scoring system again to restore the 1:1 balance. We could keep doing this until we find a sweet spot where people don't change their strategies any more.

So what I'm saying is, Ent is providing a starting point for this process. What you're providing is a recommendation for changes that we'll need to make to Ent's starting point once people begin changing their strategies.

Just to emphasize what I mean by "the macroscopic reflects the microscopic", look at Ent's numbers. They're pretty close to yours already:

North: K = 119 (kill = 3 bell flags)
East: K = 71 (kill = 2.35 bell flags)
South: K = 98 (kill = 3.3 bell flags)
West (3v3): K = 107 (kill = 3.5 bell flags)

You're suggesting we should make a kill between 2 and 3 bell flags, so Ent's numbers are almost there already. I think we should use Ent's numbers, but with slight alterations towards what you're recommending (so a bit lower than his). Like pre-empting what we think the first subsequent change to the scoring system might need to be.

The average of Ent's numbers is 98.75. How about we start off using 95 and see where that takes us?

[Tembest]

Anything below 100 is an option for me. I just think it would be logical to start where it last time was successful, the 2.5 ratio.
I kind of don't see the point of changing it to something else before we receive critic. Naturally north is a slight problem, but hey, then we'll just have a bit different
kind of a game there. 95 is acceptable. I would prefer 90 as that's already 3 in east and south.

Maybe after all it's not so bad we have different kind of sectors. Perhaps sometimes people feel like playing a flagging game, hence choose north? I don't know, but in the
end I'm not sure if the ratio should even be same in all sectors.

[Entr0py]

Kills don't matter. These are years and years of practice stats? League was different. Many star players west have MVP and most kills and bell flags.

You should just use an average of the kills in each sector. The average is 19.1305. Though I'm unsure how kills matter. It should be based on amount of players in a sector no?

That's not a weighted average. It extremely overvalues the lower number of kills in west/south in the average. If anything, the huge differences in the avg number of kills in each sector show that the number of kills is HEAVILY dependent on the sector played, even if there are some outliers where a west player such as Makron MVPs (and in fact, cases like that are included in these averages).

If that's the case, then by including the kill bonus we're also reducing the bell flag points by an identical fraction.

With what you just described, you seem to be doing: "kills + kill bonus = flags" and suggesting this is the same as: "kills = flags + kill bonus" when it's actually: "kills = flags - kill bonus".

Not quite... I'm saying
kills + kill bonus = flags * p
is equivalent to
kills = (flags + kill bonus) * q
is equivalent to
kills = flags * r; kill bonus = flags * s
where we can actually figure out the relevant values of p, q, r, and s with some work.

It looks like you're holding bell flags at 50% before reducing kills from 50% to accommodate the kill bonus.

Not quite... I'm taking a fixed quantity, bell flags (since the max bell flags points is CONSTANT, given a prac length, sector, and number of players), and then saying we want all other points (kill points + kill bonus) to be some percentage of those, p. If p = 1, then it would be equivalent to holding bell flags at 50%. But if p = 2, we'd actually be holding bell flags at 33% of the total.

Perhaps it's just easier to explain. I'd like to see something like:
Kills = 40% of all the points
Bell flags = 40% of all the points
Kill bonus = 20% of all the points (closest we can get to this number with B as an integer.)

I'll do this in all 3 ways of looking at it and show that they're equivalent.

First the way I originally did it.

Bell flags = 40% => Kills + Bonus = 60% => p = 6/4 = 1.5
Plug in p=1.5 to my earlier equations:
F*p = N*A*K + N*A*B*G/2
=>
1.5F = NAK + NABG/2
Then, it just remains to figure out K and B.
Kills = 40%, Kill Bonus = 20% => Kills = Kill Bonus * 2
NAK = (NABG/2)*2
=>
K = BG
=>
F * 1.5 = NABG + NABG/2
=>
F * 1.5 = NABG1.5
=>
F = NABG
=>
B = F / NAG
=>
K = F / NA

Next, I'll do it your way
Kills = 40% of all the points
Bell flags = 40% of all the points
Kill bonus = 20% of all the points

=> Kills = Bells; Bonus = 0.5Bells
=> NAK = F; NABG/2 = 0.5F
=> K = F/NA; B = F / NAG
Exactly the same results.

Finally, I'll lump flag bonus in with flag points
Kills = 40% => (Bell flags + Kill bonus) = 60% => q = 2/3
=>
NAK = (F + NABG/2)*(2/3)
=>
K = (F + NABG/2)*(2/3)/NA
And Bell flags = 40%, Kill bonus = 20% => Kill Bonus = Bell Flags * 0.5
=>
NABG/2 = F*0.5
=>
B = F / NAG
=>
K = (F + NA(F/NAG)G/2)*(2/3)/NA = (F + F/2)*(2/3)/NA = 1.5F * (2/3) / NA = F / NA
Again, same answer.

Without looking much into the calculations, I noticed something wrong.

Entropy, the point is NOT to distribute bell flags and kills 50 % in a prac. That's impossible. The number of kills vary a lot depending on players, teams, etc. etc.
and it even depends on the scoring system itself (how smart it is to suicide, etc.)

So, don't look at average kills in a prac. It's completely irrelevant.

It is only about not overvaluing one over another at a ding moment. In other words, the only reason we look at the value is to give specials multiple purposes.

Like leagues 1-10 if you were about to die 40 seconds prior to ding with full specials, it made sense to. Then respawn again with full specials and use those specials for
flags. That's something we want to avoid, without over-emphasising kills either. So it's not about the total proportion of bell flags, kills or anything else, but that the game
could be played in multiple ways. 6vs6 league was a perfect example of this. First time ever using a repeller for a kill actually made sense.

I'm not trying to distribute flag points vs kill points as 50-50, but trying to make it so that, on average, the two will be in some ratio to each other that we desire. Of course kills/kill points/kill bonus will always vary, but we can't even try to talk about finding a balance unless we look at the statistical averages.

Even trying to compare kill points to bell points at dings is impossible without looking at the kill averages. For example, suppose we just blindly said 1 kill should be worth 1 flag pole at a ding. Well, suppose your team averages 1 kill per player per ding. That will make kill points a completely different ratio to bell flags than if your team averages 2 kills per player per ding. It's kind of like comparing apples and oranges. We need to know how often players kill, in order to be able to try and get kill points to be some desired ratio of bell points. And how often players kill depends on the sector. Sure, changing point values might change how often players kill in the sectors, but I would argue that basing it on the lifetime prac stats gathered so far is a better gauge than just guessing or throwing out numbers.

[falconeer]

Entropy you are being vague as usual. I asked if these are practice stats, or if league is included. Years and years of stats means 10+ years. Which I doubt you have information on unless you are using league games. You only started collecting practice stat information 2 years ago when you made your bot. That doesn't qualify as years and years of stats. Also consider for every 50 east practices there is 1 west practice. So the stats are skewed, distorted, wrong, they don't account for full 12 vs 12 deathstar battles. Maybe you are treating practice as different from 12 vs 12 league, and abandoned the idea of a 12 vs 12 league. But I think we need to consider the big picture and not the sectors individually. If your stats are based on the last 2 years they are based on the same 20 shitty players that practice everyday like Eridu. Because I'm pretty sure makron hasn't been around the majority of that time, and the variety of players has dropped dramatically. I'm sorry but your idea that there are more kills in any other sector than in West, is a completely fucked and wrong idea. I hope you get my point now that I said it more strongly. Nothing influences how kills happen, people play the way they want to play. I always kill in west. A flagger who sucks at killing will avoid fighting so they dont die, and try to flag. None of this is dependent on the sector but WAY more dependent on the person and even their personality. All that matters is player count in a sector, stop trying to mathematically pin down random things like kills. This is not something you can calculate, that's why you are using averages. Just stop.

You said it yourself. You can't do it. So stop basing the formula on something you don't know.

"Of course kills/kill points/kill bonus will always vary, but we can't even try to talk about finding a balance unless we look at the statistical averages."

I'm telling you west, doesn't see less kills than other sectors. I would argue you see even more kills from west players. In the history of league, west players were always top 5 on the List. I don't know why you insist on being a retard. Kills have nothing to do with the sector you play in. You need to kill to get the bell flag no matter where you are. West has fly around and be an idiot strategies. Guess who uses those strategies? People who are idiots and can't kill. It makes sense to fly around if all you will do is die. But these strategies can be applied to south, north, and in east they come in the form of wall hiding and tubing. SECTORS DON'T AFFECT KILLS.

[falconeer]

This guy is making forumlas off of eridus stats and the 1 west practice we had this year. No wonder the bot is retarded. Average all the sectors at 19.0 or don't use kills averages at all is my vote.

[falconeer]

EFFECT*

A SECTOR DOES NOT EFFECT HOW MANY KILLS A PERSON MAKES.

And how often players kill depends on the sector.

WRONG, FUCKING WRONG, RETARDED. WRONG, STUPID, WRONG.

IT MIGHT - MIGHT - HAVE A SMALL - SMALL - AFFECT ON PEOPLE - ON PEOPLE NOT KILLS! ON PEOPLE. THUS MAKING THEM MORE LIKELY TO FOLLOW A CERTAIN STRATEGY. BUT THIS IS AN INDIVIDUAL CHOICE AND USUALLY PEOPLE PLAY HOW THEY WANT TO PLAY NO MATTER WHERE THEY ARE. EVEN IF IT MEANS LOSING. THAT'S WHY LOTS OF PEOPLE SUCK ASS AND OTHERS ARE NATURALS. The only real strategy is get the fucking flag at the bell, kill the bitch in your way, lock out the sector.