Garbage in 6P with fractional garbage and hole change rate divided by number of players is too clean

[GoogleCodeExporter]

why it is too clean?

i just realized there is some flaw here. for example in 6 players room if there 
is one garbage line added which is result or 5 people doing double (sending 1 
line) it will change hole with probability:
change-hole-rate / 5
but as this is result of 5 attacks it should have 5 chances to break, so 
actually correct probability should be in this case:
1 - (1-change-hole-rate) ^ 5

anyway calcScore function is bit messy, so somebody should review it and 
refactor if possible.

Original issue reported on code.google.com by w.kowa...@gmail.com on 19 Feb 2011 at 2:50

[GoogleCodeExporter]

thre is mistake in formula, it should say:
1 - (1-change-hole-rate/5) ^ 5

Original comment by w.kowa...@gmail.com on 19 Feb 2011 at 2:57

[GoogleCodeExporter]

ok, sorry some people don't understand me.

we are talking about change hole per attack fractional grabage. we want it to 
be like 1on1 garbage, so total number of hole changes should be divided by 
number of players -1.

but currently number of hole changes is based on number of fractional garbage 
adds not number of attack and its not always same.

consider example

6 players all do 5 doubles, each player will get 5*5 1/5th of garbage lines, 
but it will result in 5 lines of garbage added with 1 hole change but we want 
to have actually have 5 hole changes here.

Original comment by w.kowa...@gmail.com on 19 Feb 2011 at 3:34

[GoogleCodeExporter]

another thing is of course maybe instead of changing hole change probability we 
should just apply hole change every nth garbage entry. i think it was suggested 
already.

Original comment by w.kowa...@gmail.com on 23 Feb 2011 at 1:30

[GoogleCodeExporter]

In your formula:

1 - (1 - (changerate/5))^5

the two 5s are separate, right? One represents the garbage entries in that one 
line, and one represents the number of other players? Or do I have it wrong...

Hole change rate every few entries sounds like a good way to do it, we'll have 
to try that and see.

Original comment by Zirc...@gmail.com on 28 Feb 2011 at 7:27

[GoogleCodeExporter]

yes, you got it right.

Original comment by w.kowa...@gmail.com on 28 Feb 2011 at 9:39

[GoogleCodeExporter]

Wojtek and I chatted about this today. I definitely agree that his method is 
more accurate than the current system. I'll go ahead and post the idea we were 
talking about.

Imagine we have a 5 player game. All of my opponents send me a triple (two 
garbage lines) each. Under fractional rules, I'll receive a total of two 
garbage rows. Let's repeat this three more times. I receive eight lines in 
total.

Now, the goal is to make this garbage look like the garbage I would've received 
in a 1v1 game. The analogous 1v1 game in this case would be one opponent 
sending four triples. 2-2-2-2 type garbage messiness.

The conclusion we came to is that under the current circumstances, the game 
will produce 4-4 type garbage. But with his revision, it's much more like 
2-2-2-2.

That's my two cents.

Original comment by ncann...@gmail.com on 2 Mar 2011 at 9:36

[GoogleCodeExporter]

I really don't know how fractional garbage with change hole per attack should 
work. Current divide hole rate by number of players is not what we want. It's 
really hard to make it right, so i close this issue for now.

Original comment by w.kowa...@gmail.com on 13 Mar 2011 at 12:50

• Changed state: Invalid