Variance added

[Mallorea and Riva]

There you go. You had a combo of early/late movers, but even your skilled core was off by two seconds or so. Perfectly reasonable for variance.

[Ballotonia]

Ballo, if I understand it correctly, you'd added both "negative" and "positive" variance -- probably to balance out the effect of variance, so you don't expand the length of update too much. Might I recommend imbalancing the net outcome, so you're more likely to get positive variance than negative variance. If it adds 10 mins to update or something, so be it.

I'd argue it's more fun to have positive variance than negative variance -- going early into your target means you're fighting your enemy not the clock.

Sounds to me like you misunderstand the concept of variance: it's always above and below the average. See: http://en.wikipedia.org/wiki/Variance
It's mathematically impossible for variance to be only on one side of the average.

This is an indication that what you see as average is actually wrong. Suffice to say, we're not going to alter the base speed of the update just to match whichever formula you use as average speed. You'll have to adjust your formula instead. No matter what we do to the update speed, it'll have some average speed and there will be variance which stretches both above and below the average. I cannot alter the rules of math.

Ballotonia

[Unibot II]

Ballo, if I understand it correctly, you'd added both "negative" and "positive" variance -- probably to balance out the effect of variance, so you don't expand the length of update too much. Might I recommend imbalancing the net outcome, so you're more likely to get positive variance than negative variance. If it adds 10 mins to update or something, so be it.

I'd argue it's more fun to have positive variance than negative variance -- going early into your target means you're fighting your enemy not the clock.

Sounds to me like you misunderstand the concept of variance: it's always above and below the average. See: http://en.wikipedia.org/wiki/Variance
It's mathematically impossible for variance to be only on one side of the average.

This is an indication that what you see as average is actually wrong. Suffice to say, we're not going to alter the base speed of the update just to match whichever formula you use as average speed. You'll have to adjust your formula instead. No matter what we do to the update speed, it'll have some average speed and there will be variance which stretches both above and below the average. I cannot alter the rules of math.

Ballotonia

No, what I'm suggesting is you have a base speed and you're slowing it down and speeding it up so the end result is relatively even as though you ran the update as just the base speed.

I'm asking for it to slow down more often than it speeds up.

I ask this because its more fun for both sides if either side moves early as opposed to late.

[Ballotonia]

No, what I'm suggesting is you have a base speed and you're slowing it down and speeding it up so the end result is relatively even as though you ran the update as just the base speed.

I'm asking for it to slow down more often than it speeds up.

I ask this because its more fun for both sides if either side moves early as opposed to late.

You already have that now. It's just a different base speed than what you think there is. You could also imagine there's a really slow base speed, and that the server only speeds up to various degrees. It's all the same thing, since we don't advertise the actual base speed.

Ballotonia

[Unibot II]

No, what I'm suggesting is you have a base speed and you're slowing it down and speeding it up so the end result is relatively even as though you ran the update as just the base speed.

I'm asking for it to slow down more often than it speeds up.

I ask this because its more fun for both sides if either side moves early as opposed to late.

You already have that now. It's just a different base speed than what you think there is. You could also imagine there's a really slow base speed, and that the server only speeds up to various degrees. It's all the same thing, since we don't advertise the actual base speed.

Ballotonia

Then why is update nearly always quicker than when traditional triggers predict them to be? And update is the same speed practically?

(This may not be relevant, since I think I figured out a new method for triggering that is perfectly accurate. >_>)

[Mallorea and Riva]

You already have that now. It's just a different base speed than what you think there is. You could also imagine there's a really slow base speed, and that the server only speeds up to various degrees. It's all the same thing, since we don't advertise the actual base speed.

Ballotonia

Then why is update nearly always quicker than when traditional triggers predict them to be? And update is the same speed practically?

(This may not be relevant, since I think I figured out a new method for triggering that is perfectly accurate. >_>)

Wouldn't a triggering method that exists in a world of variance be inherently flawed? The old triggers we used were perfectly accurate. And maybe because you're unlucky to answer your question to Ballo?

[Unibot II]

Then why is update nearly always quicker than when traditional triggers predict them to be? And update is the same speed practically?

(This may not be relevant, since I think I figured out a new method for triggering that is perfectly accurate. >_>)

Wouldn't a triggering method that exists in a world of variance be inherently flawed? The old triggers we used were perfectly accurate. And maybe because you're unlucky to answer your question to Ballo?

I've tried it 10 times, been 100% accurate, 10 times. So I need to test it more.. but I suspect you'll see it out on the field soon.

[Mallorea and Riva]

Wouldn't a triggering method that exists in a world of variance be inherently flawed? The old triggers we used were perfectly accurate. And maybe because you're unlucky to answer your question to Ballo?

I've tried it 10 times, been 100% accurate, 10 times. So I need to test it more.. but I suspect you'll see it out on the field soon.

That rather defeats the purpose of variance being added. That having been said, I'll believe it when I see it

[General Halcones]

I have noticed a significant increase in the update length as a result of the added variance.

The two updates are now 5-6 minutes longer than they were before the variance was added. Not only that, but there are increasing lag problems through the update which Uni did previously mention.

I believe it may need tweaking to avoid the update lengthening. Having said that, I don't really mind - it of course gives us more time to hit targets. Just an observation.

[Unibot III]

I have noticed a significant increase in the update length as a result of the added variance.

The two updates are now 5-6 minutes longer than they were before the variance was added. Not only that, but there are increasing lag problems through the update which Uni did previously mention.

I believe it may need tweaking to avoid the update lengthening. Having said that, I don't really mind - it of course gives us more time to hit targets. Just an observation.

You must have been away when the old server bloomed to three hours an update and beyond.

[General Halcones]

Yes indeed. I only remember it nearing two hours, and remember how much TITO complained about it. 3 hours would have been very interesting indeed.

The only reason I bring this up is because I believe it is due to the extra variance - it is not supposed to be lengthening the update. I can see how a random function could lead to longer updates somedays and shorter updates on others, however, if the end time of update is shifting by over 5 minutes, then clearly something needs to be tweaked.

[Ballotonia]

When the variance was added there was no easily measurable effect on the length of the update (although I can tell you exactly what the effect is, but won't). My log of data dumps shows total region update completions at 66-68 minutes, comparable to the way it was before, and a slow increase on par with an increase in how much there was to update. Only this morning it was 72 minutes. That's an aberration. The thing which has changed since yesterday is a major update of a forum server. There may be sync issues...

NB: timing info I have at hand is only for the major update, I got no data like that for the minor. Are you perhaps only talking about the minor update?!?

Ballotonia

[[violet]]

That's an aberration. The thing which has changed since yesterday is a major update of a forum server. There may be sync issues...

Yep, highly likely to be a short-term thing related to forum server changes.

[General Halcones]

I see, thanks for that. I was referring to both updates actually. Last one just finished later than usual again, maybe 72 minutes long which is longer than usual. I didn't expect the forum server could have any effect, but perhaps it can. Is it possible the new forum server could be causing even more variance, in addition to the deliberate variance? It sure seems to have increased even more.

[Sichuan Pepper]

Does activity have an effect on variance? If taggers have a team out of 5 or 6 and are moving and switching often, causing defenders to be hitting the reports page often...will this jigger the update slightly?