Terrible algorithms

Forum rules
Please DO NOT POST BUGS on this forum. Please report (and vote) bugs on : https://boardgamearena.com/#!bugs
User avatar
Stealthpup
Posts: 41
Joined: 26 November 2019, 02:36

Re: Terrible algorithms

Post by Stealthpup »

Just as a datapoint that keeps getting weirder and weirder for me:

My last 9 games in a row, I've gotten fewer doubles than my opponent.

In the last 4, the doubles differential has been -5, -5, -3, and -5.

I understand statistical anomalies, but this is pretty far out there.
User avatar
RicardoRix
Posts: 2106
Joined: 29 April 2012, 23:43

Re: Terrible algorithms

Post by RicardoRix »

It's most noticeable when you throw the dice and the results are not what you had hoped for.

After a complete game, I can only conclude that the algorithm is totally broken and the dice are completely random.
User avatar
Romain672
Posts: 1014
Joined: 05 April 2016, 13:53

Re: Terrible algorithms

Post by Romain672 »

Stealthpup wrote: 16 December 2020, 18:41 Just as a datapoint that keeps getting weirder and weirder for me:

My last 9 games in a row, I've gotten fewer doubles than my opponent.

In the last 4, the doubles differential has been -5, -5, -3, and -5.

I understand statistical anomalies, but this is pretty far out there.
Ok let's see.

How can I convert your post into something likely...
Probabilities to have -3 (or less) four times in a row for yourself is 14.2%^4=0.04%. 1/2500. You did 121 games, I agree it's unlikely.

If I do 8x20 rolls, and sum the differential to find -18 or less, I find 0.0078%. (1/12.676)
If I increase the number of rolls to 25: 8x25, I got 0.0127%. (1/7.897)


All these value are between 2k and 20k.
I suppose it's strange for a specific player, but normal for a community.

Just continue playing, this streak of -3/-5 will normally stop really soon.


And never forget the probability to find something strange is I believe 10x more likely if you count: the opposite (you who got more doubles than the other player), a specific roll in a row (for both, you, or the opponent), your number of double, opponent number of double, and lots of more unlikely event (123456, doubles in a row, or something more related to backgammon than random generation like staying blocked 10 turns in a row).
User avatar
Stealthpup
Posts: 41
Joined: 26 November 2019, 02:36

Re: Terrible algorithms

Post by Stealthpup »

I was too lazy to work out the actual probabilities -- I thank you for your efforts.

And I'm aware that one notices anomalies far more than "regular programming".

I will be happy when this particular anomaly smooths out -- I'll never get to Elite at this current rate.

I do like to joke that I warp probabilities around me -- I remember a game of Amoeba Wars that I lost when my 23-ship armada was decimated by a single scout; we calculated the odds of that happening at about 50,000 to 1.
User avatar
Romain672
Posts: 1014
Joined: 05 April 2016, 13:53

Re: Terrible algorithms

Post by Romain672 »

I really like to calculate probabilities. Yours was pretty high compared to some others.

But usually i'm stuck to some thousands. It's rare I reach 10k+.
User avatar
RicardoRix
Posts: 2106
Joined: 29 April 2012, 23:43

Re: Terrible algorithms

Post by RicardoRix »

look at it the other way.

How far does the programmer have to go in order to make 5 games in a run just for YOU biased ?
And why has it specifically picked on you.
And then think how difficult (neigh on impossible) and if it's worth their time.

Also, the source code for this game is freely available.
User avatar
Stealthpup
Posts: 41
Joined: 26 November 2019, 02:36

Re: Terrible algorithms

Post by Stealthpup »

And the bad streak finally breaks!

Got a +2 doubles differential, and finally squeaked out a win.
User avatar
Giordano Bruno 1974
Posts: 225
Joined: 28 April 2020, 19:56

Re: Terrible algorithms

Post by Giordano Bruno 1974 »

RicardoRix wrote: 16 December 2020, 21:01 look at it the other way.

How far does the programmer have to go in order to make 5 games in a run just for YOU biased ?
And why has it specifically picked on you.
And then think how difficult (neigh on impossible) and if it's worth their time.

Also, the source code for this game is freely available.
To be fair to stealthpup he / she wasn't complaining about how the random function was broken. S/he was remarking a series of unusual dice throws.
User avatar
0n1
Posts: 6
Joined: 06 December 2020, 19:47

Re: Terrible algorithms

Post by 0n1 »

What random function BGA uses and how it's implemented in the Backgammon?
User avatar
RicardoRix
Posts: 2106
Joined: 29 April 2012, 23:43

Re: Terrible algorithms

Post by RicardoRix »

Shouldn't that have been posted under the title of 'Unbelievable bad luck' thread?
Last edited by RicardoRix on 17 December 2020, 14:05, edited 1 time in total.
Post Reply

Return to “Backgammon”