I’ve been playing for a couple of weeks without outrageous dice results. Swingy but like a regular, in-person A&A game. Today, however, it’s been really bad. Super one-sided results.
I know that Beamdog ran experiments on PRNG algorithms last year at some point. Is that happening again right now?
This is how randomness works. It gets perceived by us as not that random. Standard human cognitive biais.
@Nosho said in PRNG experiment going on right now?:
This is how randomness works. It gets perceived by us as not that random. Standard human cognitive biais.
dismissing things as standard human cognitive bias without data or evidence
hm
i feel that might be standard human cognitive bias
https://en.wikipedia.org/wiki/Cognitive_bias
omg it is!
@aardvarkpepper Please provide evidence that PRNG dice are significantly deviating from random results. Until then, anecdotal “evidence” should indeed be dismissed as standard human biaises in the face of randomness.
@Nosho said in PRNG experiment going on right now?:
@aardvarkpepper Please provide evidence that PRNG dice are significantly deviating from random results. Until then, anecdotal “evidence” should indeed be dismissed as standard human biaises in the face of randomness.
Since you are the one dismissing others’ claims, I find it reasonable that you provide evidence that PRNG dice are not significantly deviating from random results.
No?
. . . because it’s not your job?
So it would be the developers’ job then?
And if it is the developers’ job - should we take the harsh attitude that the developers ought to have foreseen the issue and taken preventative measures against every conceivable issue in the first place? Or should we extend the benefit of the doubt and say that the developers couldn’t reasonably foresee every issue? Then wouldn’t it be reasonable that players should report perceived issues? And perhaps even that the developers take reports seriously?
The burden of proof should be on the reporting player? Do you really find it reasonable that every player that does report issues should be held responsible for amassing and writing up thousands of hours of work worth of data collection and analysis - or - taking 1942 Online apart and inserting bits of code, not only on their own system, but thousands of users’ systems? Because that’s the only way you’re going to get the data that you need. That sounds like a proper job to me. The sort of job you get paid for.
But if you really find it reasonable that others put in loads of unpaid work, I think it’s reasonable that we can expect you to put in loads of unpaid work. So let’s start with your defining the nature of the issue and listing specific steps as to how that issue may be addressed.
And let’s not use “human bias”, yes? You need mathematics and real analysis to address the issue. If you simply dismiss opinions you disagree with as “standard human bias” absent any real evidence, that is itself bias, and you negate your own argument.
@aardvarkpepper Most PNRG are widely considered to be a valid approximation of randomness, so yes, people claiming differently have the burden of proof. I don’t need data to assume a modern PRNG is extremely close to random.
@Nosho said in PRNG experiment going on right now?:
I don’t need data
It’s true that you don’t need data to make subjective claims.
But really, everyone needs Data.
Didn’t mean to start a thing. Was just wondering if other suspected the same. But Maybe @Nosho is right, just my incorrect perception based on limited sampling.
@aardvarkpepper Very intellectually honest of you to omit the rest of my sentence: “I don’t need data to assume a modern PRNG is extremely close to random”
If you or anyone else want to convince anyone that their anecdotal evidence is relevant to supporting the claim that PRNG has significant and noticeable patterns/trends that make it substantially different from randomness, then you are the one that has the burden of proof. The people that assume that PRNG is a valid randomness generator can be confident in their assumption based on the fact that, for example, GPS, works pretty well. When you make a claim that goes against the commonly accepted current understanding of things, you back it up and submit it to others for scrutiny, or hold your peace.
@djensen said in PRNG experiment going on right now?:
Didn’t mean to start a thing. Was just wondering if other suspected the same. But Maybe @Nosho is right, just my incorrect perception based on limited sampling.
After losing 5 out of 6 bombers on a SBR and seeing my surviving bomber score a 1 in a recent game, I have to admit it was very tempting for me to believe something weird was going on there. But given the fast pace and the sheer amount of games we can play online, weird stuff is bound to happen often, even due to pure randomness. I estimate I’ve conducted over 2000 SBRs on A&A Online so far. Some extreme results are to be expected at some point, and will be very salient to our memory. Cheers
@Nosho You should start manually recording your SBR results. :rolling_on_the_floor_laughing:
You can dictate terms to me when you pay me. Until then -
PRNG evaluation is designed to evaluate groups of single independent events such as dice rolls. But Axis and Allies dice are properly evaluated as groups of groups of independent events.
@aardvarkpepper :thinking_face:
@aardvarkpepper So is it your actual belief that the PNRG dice in A&A Online are significantly deviating from a random distribution? If so, in what way do you think they are deviating from a random distribution, and what is your evidence supporting this opinion?
I locked the thread because I know where this is going - an explosive reponse. If you two would like to chat privately about it, go ahead. Keep in mind forum rules still apply to chats.
