Jump to content
punkUser

I tested my dice...

Recommended Posts

27 minutes ago, Tlfj200 said:

I love that you'd rather argue semantics than just accept how the rules work, and that you know work, and you know will actually be in force at worlds.

Are you saying that you'd tell a worlds marshall, Frank brooks or Max Brooks, about the nuances of english language and that your opponent cannot inspect your damage deck or share dice?

A Marshal can mandate it. at that point, it wouldn't be a request.

On a personal note, I doubt I would ever ask to share dice. I also would never say no if requested to. If the request is presented, my opponent has a reason to do so. I want to have fun while playing this game. If making my opponent more comfortable regarding dice helps us both have fun playing, I'm all for that.

Share this post


Link to post
Share on other sites
Posted (edited)
34 minutes ago, theBitterFig said:

In truth, correct.

Practically speaking, maybe not.  Under one interpretive scheme, the requesting player would have to hassle not just any judge, but the full marshal.  This is probably untenable in a larger tournament, and essentially eliminates the rule entirely.

Getting a marshal isn’t a hassle at all, especially since a judge is usually called first for disputes, and a player refusing to share dice will absolutely get a marshal’s attention.

Edited by miguelj

Share this post


Link to post
Share on other sites
1 minute ago, miguelj said:

Getting a marshal isn’t a hassle at all, especially since a judge is usually called first for disputes, and a player refusing to share dice will absolutely get a marshal’s attention.

Could be.

Haven't played too many really large events, and to some extent am just guessing about very large (150+) events.

Share this post


Link to post
Share on other sites
1 minute ago, theBitterFig said:

Could be.

Haven't played too many really large events, and to some extent am just guessing about very large (150+) events.

Understandable, but yeah, getting the marshall isn't really a big deal.

I'd really like to normalize the idea of simply calling judges and/or marshals as a practice for ambiguity or tough situations. It's what they're there for, and really isn't a bother to them.

Share this post


Link to post
Share on other sites
1 minute ago, Tlfj200 said:

Understandable, but yeah, getting the marshall isn't really a big deal.

I'd really like to normalize the idea of simply calling judges and/or marshals as a practice for ambiguity or tough situations. It's what they're there for, and really isn't a bother to them.

Flip side is that at a small event, TO usually plays.  Getting them over for an ambiguous call interrupts their game too.  Doesn't seem too hard to get into a habit from one size of event, which doesn't necessarily apply in others.

Share this post


Link to post
Share on other sites
13 hours ago, gadwag said:

This is great.  I'm happy seeing @punkUser get props in the hacker community.  Might inspire similar projects.

Also, reading the article comments: Gee, I'm wondering if the person making this comment is just BSing to sew dissent or take swipes at our community.  Kinda seems like he didn't read the article, he just scanned it (at most) and started ranting.

Quote
Wil says:

It’s of note that he actually finds an approximately 95% similarity to perfect abstract fairness. Despite writing a 60+ page report, he’s fumbled the basics of statistics, and ignored confidence levels. His own study disprovies his conclusion. There is no benefit to be gained from the minor variation within these dice, and claiming that there is is dishonest and causing “panic” in an already reactionary and hysterical community

I like the word "disprovies".  It carries all the weight of "The Resisties!" from Invader Zim.  :D

Share this post


Link to post
Share on other sites

BTW, I'm serious that multiple people should hit up FFG with the same question about whether sharing dice can be refused.  FFG ignores me because I'm personally boring (harsh but true!), and also only one person.  If more people ask, that puts it on the desk of someone who matters faster (or at all).  Use this link to get to the form.

Share this post


Link to post
Share on other sites
Posted (edited)
14 hours ago, ScruffyNerdHerder said:

Not to detract from the discussion about what is and is not cheating/legal ...

Please do actually ;)

Quote

... but have you tried using your ML model to recognize multiple results in a view field?  I’m not really sure what the application would be but maybe there could be something cool done with interpreting results during streams from raw dice box footage.

Yes, definitely possible, just a little more complicated to have variable numbers of outputs. As an aside when testing some different dice that were close to the color of the wood (and thus I couldn't easily crop them out by hue), I threw the "entire compartment" image at it and let it learn how to find the die in the box as well. That definitely works as well, but it needs quite a bit more training data, as the network is necessarily larger (particularly if you include rotations of the input data).

Quote

If you have attack dice that are positively biased by 5 percent the expected damage of 1 die is 0.8 if you have a focus result. With  howlrunner in play that goes up to 0.96.  Assuming everyone gets to shoot at range 2 that nets out to 14*0.96, or 13.44 damage results vs 14 * 0.75 (12.25)  (I know these estimates are high, but as a learned man on mobile I don’t have access to the binomial distribution function right now). 1.2 extra damage per round seems like a pretty distinct advantage and also seems likely to snowball (the more hits you roll over time the more likely it is that your opponent are rolling fewer red dice)

There's a discussion of the impact in one of the sections and yes, 5% is pretty clearly too far off. I believe the example I gave was that on 3 defense dice it's closer to having/not having a focus than the expected distribution. Most folks would have a problem with that sort of passive effect being active on every roll :P

11 minutes ago, Wazat said:

Also, reading the article comments: Gee, I'm wondering if the person making this comment is just BSing to sew dissent or take swipes at our community.  Kinda seems like he didn't read the article, he just scanned it (at most) and started ranting.

Right, I guess he skipped the aforementioned section at the very least. Very probably he doesn't play X-Wing as well, as with the above example I don't know anyone who would think those dice are "fine".

13 hours ago, Hiemfire said:

If a request cannot be refused it is not a request.

Not to fuel the fire further, but I request many things of my kids every day that cannot be refused :P

In all seriousness I agree with @Brunas here... there's no reason for that section to be in there if it isn't meant to be "by default, mandatory". As noted, I can and do request all sorts of things from my opponents over the course of a game and hardly need the regulations to spell that out for me. The deference to marshal being the final authority on everything is exactly that: of COURSE the marshal gets to make the final call, especially if one or both players are being obtuse about it.

Any further reading into it than that is special pleading IMO... Furthermore, does anyone actually know a marshall at any decently-sized event who would not rule it as mandatory once requested unless there were some exceptional circumstances? I certainly don't. It should be very clear why the rules are there are this point to everyone, right?

Edited by punkUser

Share this post


Link to post
Share on other sites

Here are 2 more interesting reads on dice testing, albeit with numbered dice. The first one is a shorter summary of the second longer article.

(one): https://arstechnica.com/gaming/2016/08/how-fair-is-your-d20/

(two): http://www.markfickett.com/stuff/artPage.php?id=389

They are both interesting reads, and they agree with the initial google document that spawned this thread.

Share this post


Link to post
Share on other sites

 

5 minutes ago, punkUser said:

... but have you tried using your ML model to recognize multiple results in a view field?  I’m not really sure what the application would be but maybe there could be something cool done with interpreting results during streams from raw dice box footage.

There was a recent talk or video by google where they do some live tracking of objects for fairs, and they retrain on location with the new lighting conditions, likely sending the data to their servers or bringing some expensive graphics card for that. Just to say that changing lighting conditions should not be underestimated! The dice recognition here had fixed lighting conditions with the ocational cat shadow maybe... if punkUser has one idk.

 

Also, the raw data (individual rolls) would be cool to have, for critics like on that hacker article comment (so rude) and for myself so I can redo some of the analysis to finally understand chi-square :D Being xwing it gives me motivation to finally do that after work! If it's there already I totally missed it twice, would be classic me.

Share this post


Link to post
Share on other sites
Posted (edited)
19 minutes ago, RunnerAZ said:

Here are 2 more interesting reads on dice testing, albeit with numbered dice. The first one is a shorter summary of the second longer article.

(one): https://arstechnica.com/gaming/2016/08/how-fair-is-your-d20/

(two): http://www.markfickett.com/stuff/artPage.php?id=389

They are both interesting reads, and they agree with the initial google document that spawned this thread.

Thanks! Note that there are a few more linked in the article itself, as well as some articles that discuss dice manufacturing processes that are an interesting skim. 

There are a few other folks who have made automated dice rollers over the years. The Ars article references this test which is probably the most similar to what I did, and definitely worth a read regardless:
http://www.markfickett.com/stuff/artPage.php?id=389

Edited by punkUser

Share this post


Link to post
Share on other sites
3 minutes ago, punkUser said:

Not to fuel the fire further, but I request many things of my kids every day that cannot be refused

Then you aren't requesting. You're telling while trying to be polite about it. When I request my that kid does something, it is with the expectation that refusal will be respected. When I instruct them to do something they don't have the option to refuse.

Respecting your opponents' right to refuse a request is the difference between "Fly Casual" and being a dictatorial *** (and I'm seeing allot of the fear/concern induced latter in this thread). The Marshal still has the authority to overrule either the request itself or a refusal to comply to the request. Just because language has gotten loose in application does not mean the foundational meaning of "request" has changed.

@Tlfj200 If FFG's intention is for it (anything that the tourney regs have with a "may/can request" attached to it) to be mandatory then they need to reword it.

Share this post


Link to post
Share on other sites
4 minutes ago, Xeletor said:

There was a recent talk or video by google where they do some live tracking of objects for fairs, and they retrain on location with the new lighting conditions, likely sending the data to their servers or bringing some expensive graphics card for that. Just to say that changing lighting conditions should not be underestimated! The dice recognition here had fixed lighting conditions with the ocational cat shadow maybe... if punkUser has one idk.

Yep, almost anything is possible with machine learning, it's just a question of how much time, computation power and most importantly training data you have :)

I tried to get the lighting fairly consistent but it varied a bunch over the day (was not a completely dark room) and it didn't turn out to be a big deal. Indeed I intentionally varying the colors/lighting a bit procedurally during training to reduce over-fitting on those things.

4 minutes ago, Xeletor said:

Also, the raw data (individual rolls) would be cool to have, for critics like on that hacker article comment (so rude) and for myself so I can redo some of the analysis to finally understand chi-square :D Being xwing it gives me motivation to finally do that after work! If it's there already I totally missed it twice, would be classic me.

I'll likely be making the repository with the code, training data and results open source in the next few days, so feel free to have at it and post back here :)

Share this post


Link to post
Share on other sites
5 minutes ago, Hiemfire said:

Then you aren't requesting. You're telling while trying to be polite about it. When I request my that kid does something, it is with the expectation that refusal will be respected. When I instruct them to do something they don't have the option to refuse.

Do you really find this semantic distinction you're trying to make a stimulating conversation here? People are politely pointing out that it's a really poor argument about the intentions of the tournament regulations when weighed against its presence in the regulations in the first place, but if you want that spelled out there you go :)
 

Respectfully you've said your bit on this now and anyone who was going to be convinced has been. Can we all perhaps take the "tournament RAW" discussion to another thread at this point to leave this one open for actual discussions of the article and impacts here? I'm fine with continuing to discuss whether dice sharing itself makes sense as that is one of the recommendations in the article, but I don't think arguing semantics of the current tournament rules is appropriate.

Share this post


Link to post
Share on other sites

@punkUser I agree that dice sharing does make sense to mitigate any inconsistencies in variance within a match and as a counter to rigging the dice, which I would consider proactively choosing dice to slant the variance to a preferred result. There was a suggestion for a fair way of doing it a few pages back (one person requests, the person asked chooses which set becomes the shared pool) that I see merit in. The side argument is looking more and more to be a conflict of core ideologies, so I'm dropping it.

Share this post


Link to post
Share on other sites
25 minutes ago, Hiemfire said:

Then you aren't requesting. You're telling while trying to be polite about it. When I request my that kid does something, it is with the expectation that refusal will be respected. When I instruct them to do something they don't have the option to refuse.

Respecting your opponents' right to refuse a request is the difference between "Fly Casual" and being a dictatorial *** (and I'm seeing allot of the fear/concern induced latter in this thread). The Marshal still has the authority to overrule either the request itself or a refusal to comply to the request. Just because language has gotten loose in application does not mean the foundational meaning of "request" has changed.

@Tlfj200 If FFG's intention is for it (anything that the tourney regs have with a "may/can request" attached to it) to be mandatory then they need to reword it.

I mean, that's where we disagree.

Most do not need that to be reworded. That's sort of the point.

It seems like an argument in bad faith, because if your idea is to be believed, they literally wrote extra words for no reason. Literally them pointing out "players are allowed to breath" would be as useful, as it would be writing words for the sake of words.

Share this post


Link to post
Share on other sites
5 minutes ago, Hiemfire said:

@punkUser There was a suggestion for a fair way of doing it a few pages back (one person requests, the person asked chooses which set becomes the shared pool) that I see merit in.

Right, I've been asking TOs for the past few months and that is how all of them so far have said it works. I certainly am happy to let my opponent pick which dice to use because it doesn't really matter. The TOs have said if both players can't agree for some reason they are generally happy to provide house dice as well.

Share this post


Link to post
Share on other sites
Posted (edited)
13 minutes ago, RunnerAZ said:
Be careful!
Image result for terminator

I mean we all knew it was going to start with X-Wing ultimately, right? What else would motivate such a level of distaste for the human race? ;)

Kidding aside, X-Wing community is awesome, love you all :)

Edited by punkUser

Share this post


Link to post
Share on other sites
Posted (edited)

Uh oh... My Wullffwarro War Crimes fleet may be a bad influence on the AI.  Especially if it becomes more viable than "casual-only" after the points update.  I need to start flying even more janky fleets to throw off the learning algorithm!  (and I should be naming my fleets differently)

Also, please no one accept a game of "Thermo-Nuclear X-Wing".  The only winning move is not to play.

 

IMO in a shared pool I prefer a coin flip to determine which player gives attack dice, and which player gives defense dice.  That makes it harder for a player to benefit from bringing bad evade dice (when they're flying a low-agility fleet), as they only have a 50% chance of foisting it upon the imperial aces player.

There's still ways to game the system when sharing a dice pool, but it's still superior to not sharing.

Edited by Wazat

Share this post


Link to post
Share on other sites
Posted (edited)

Ok, so I've been watching this over the last few days and I just want to make a few comments. Agree/disagree/comment/freak out as you will.

First of all, not to discount @punkUser's excellent work, but I really don't find this all that worrying, and here's why: yes he found that the majority of dice have some level of variant bias off of the norm, but only a few were so far off to count as 'rigged'.  The other salient point here is that they aren't all rolling one particular way. Some roll more blanks, some more eyes, some more hits/crits/evades. So if you roll dice like almost everyone I've played with (just grabbing the number needed for a particular roll out of a pile of 10-20 dice) aren't those differently weighted dice basically going to even each other out over time, at least insofar as 95% of casual or store/monthly tournaments matter?

Sure, there's the possibility for abuse, but how many people are actually going to take the time to test their dice to do this? And if you're worried about your dice being bad, find someone like me who has waaay too many base dice anyway and ask them to trade randomly. I know I'd happily do it. This should even out variance among the community.

I think if you're in the top cut of a regional tournament or above, sure sharing dice makes sense, otherwise this seems like a lot of worry over some pretty minor issues.

Edited by Dr Moneypants

Share this post


Link to post
Share on other sites
38 minutes ago, Dr Moneypants said:

Ok, so I've been watching this over the last few days and I just want to make a few comments. Agree/disagree/comment/freak out as you will.

First of all, not to discount @punkUser's excellent work, but I really don't find this all that worrying, and here's why: yes he found that the majority of dice have some level of variant bias off of the norm, but only a few were so far off to count as 'rigged'.  The other salient point here is that they aren't all rolling one particular way. Some roll more blanks, some more eyes, some more hits/crits/evades. So if you roll dice like almost everyone I've played with (just grabbing the number needed for a particular roll out of a pile of 10-20 dice) aren't those differently weighted dice basically going to even each other out over time, at least insofar as 95% of casual or store/monthly tournaments matter?

Sure, there's the possibility for abuse, but how many people are actually going to take the time to test their dice to do this? And if you're worried about your dice being bad, find someone like me who has waaay too many base dice anyway and ask them to trade randomly. I know I'd happily do it. This should even out variance among the community.

I think if you're in the top cut of a regional tournament or above, sure sharing dice makes sense, otherwise this seems like a lot of worry over some pretty minor issues.

The general conclusion is yes, the dice are not a problem unless someone deliberately builds a pool of good dice by doing a bunch of testing.  If they do that, they're not going to be willing to randomly exchange dice: they put a ton of work into getting that competitive advantage.  Someone in this thread has already refused to share dice pools because they want to keep that advantage.  So people can hedge against their own bad dice by trading around, but that only solves your dice.

It's the people who are willing to put in the work to make a victory dice pool that are the problem.  With prizes, fame, and even just competitive ego on the line, people will do it.  Not everyone, but it just takes a few.

The variation of any common dice pool is going to be very muted, but a well-assembled victory dice pool can roll so well it'll be like having a free lock/focus on every roll.  That's the concern.

 

@punkUser: I wonder if 1st edition dice are better or worse than 2nd edition dice.  Looking mine over, they're a different color (possibly color loss due to age) and some are maybe not as sharp-edged as the 2E dice (hard to tell).  It's possible the company making the dice changed its methods over time to produce a better product, or maybe a cheaper/worse product.  :)

Share this post


Link to post
Share on other sites
1 minute ago, Wazat said:

Looking mine over, they're a different color (possibly color loss due to age) and some are maybe not as sharp-edged as the 2E dice (hard to tell).

I also wonder if/how age, wear and tear all come into it. All in all, it's an interesting topic and study.  

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


×
×
  • Create New...