In response to your incessant complaining about the "obvious unfairness" about the ticket process, I asked for a reply on the equations used on the transforma mailing list and "demanded" they show us the answer. BTW if you or anyone else really cares what's going on it's NOT going on here on tribe, it's on that list. Here is the response I received:
-----------------------------
- Issue a statment on how tickets were decided this year, including how randomness was determined, AND how the remaining tickets will be alloted.
I have absolutely no comment on any of the other things in your email, neither for nor against, but I would like to point out that I think this point has already been answered.
On March 19th, WordPlay wrote:
---
While not comfortable with supplying the computer code on a public venue, the randomization process was based on a 20 digit alpha-numeric sequence that was provided by a random officemate, with no connection to Transformus, of one of our BOD members. This number was then applied by computer to the list of requests, which again were identified by a number rather than a name, until 1350 invites were fulfilled. No consideration was given to whether one or two invites were requested.
---
I have done some computer coding myself, so I can add some information here. Pretty much all computer programming languages have a function that is capable of spitting out a stream of pseudo-random numbers. These functions are used in all sorts of cases where a program needs to emulate a random event, like rolling a die or picking numbers out of a hat or what have you.
The "random" function can be used to generate a pseudo-random number between 1 and an arbitrary maximum. So if you had 2,000 ticket requests and you needed to randomly choose 1,350 of them, you would assign each of the requests a number between 1 and 2,000, then you would have the "random" function spit out 1,350 numbers between 1 and 2,000. Each of the numbers that came up would be a ticket request that got fulfilled.
The "random" function is called pseudo-random, not random, and this sometimes leads lay-people to conclude that it is in some way inferior to a random function, but it's really not--not for most intents and purposes. In order to generate TRUE randomness, you need an external source of information. For example, a certain company makes money by generating TRULY random streams of digits, using the background microwave radiation of the sky as a source. Most of us don't need to go to that much trouble for little things like simulating die rolls or picking names from a hat.
Pseudo-random number generation algorithms generate streams of digits that are highly mathematically correlated to actually random events. They are a very close approximation of a truly random stream of digits. A lot of research has gone into this field, because random numbers are important in areas like cryptography, not just rolling dice and picking names from a hat. If a random number generation algorithm is used for cryptography, and it turns out not to actually be random, it compromises the encryption, which is a Very Bad Thing, so the random-number-generation algorithms in popular computer programming languages are well-vetted. Every so often, a popular random-number-generation algorithm is found to be flawed, but really I think it's quite unlikely that this affected the Transformus ticket process.
One difference between a truly random event and a pseudo-random event is that the truly random event is not predictably repeatable, whereas a pseudo-random event is. The way this works is that the "random" function takes a number as a "seed". That number is used to mathematically generate the first digit of the pseudo-random number stream, and then the first digit is fed back in as the new "seed" to generate the second digit, and so on. You can therefore see that, if you enter the same "seed," you will get the same stream out each time. The stream will have the same statistical distribution as a randomly-generated stream, but it's actually totally repeatable.
It is true that the size of the seed limits the number of possible random streams, so the smaller the seed, the smaller the number of unique number streams that can exist. The seed, according to WordPlay, was a 20-digit alphanumeric string. I'm guessing it was hexadecimal (base-16) since he said alphanumeric and not numeric. This means that there were 18,446,744,073,709,551,615 possible unique streams of digits, which is plenty, I'm thinking, especially given that Transformus only pulled the first 1,350 numbers off the front of the stream. (I got that number by converting FFFFFFFFFFFFFFFFFFFF hex into decimal.)
Could the repeatability of the pseudo-random system be used to cheat the system? Sure. If you knew ahead of time what seed was going to be used, you would know exactly which 1,350 numbers were going to come up, and you could make sure that the people you wanted to get invites were assigned those numbers. This is why WordPlay points out that the seed was selected by a neutral third party and nobody associated with Transformus knew what it was going to be ahead of time.
This system is also inherently auditable. The organizers could, if they wanted to, show the database of ticket requests, including which number was assigned to each ticket request, then they could reproduce the generation of pseudo-random numbers, using the same seed that was originally used. It could then be confirmed that the ticket requests that were granted correspond to the numbers that the pseudo-random function generated. It would be very difficult to fake this out, as long as the "random" function was intact. And if the "random" function was not intact--that is, if it had been manipulated to give ticket requests to certain people and not to others--then this would be easy for a mathematician to detect.
Try Me!
--------------------------------
The process I was charged with devising and executing to fairly turn away 844 requested invites is provably fair and could be audited by someone qualified to evaluate it in good faith. I never understood the "20-digit alphanumeric sequence" gibberish. I asked Debra to ask someone unrelated to Transformus to give her a large, positive whole number, to send it to her via e-mail, which Debra then sent to me. The point of doing that was to provide a seed value for the MySQL RAND() function so that the SQL SELECT statement would have results that were reproducible, and yet the exact ordering of the outcome was arms-length from me, so that no one could say that I sought a seed value that happened to produce results I preferred. I actually cared about fairness, auditability and transparency to that degree. (About me briefly: I received honors in computer science from Oberlin College in 1987, invented a software product and started a software company in 1996 and sold part of it in 2002, thereby allowing me the free time to devote hundreds of hours of volunteer work, for no compensation, trying to help this little regional burn be what it advertised itself to be.) Turns out, my gifts and efforts at fairness were being perversely undermined by very ill-intentioned people, forcing me to withdraw my work so that they would not be used as cover for deeply dishonest activity that was going to turn the burn into a sick joke of a frat party masquerading as something that lived up to the stated principles.
Sorry that those of you who are unrelated to this dishonesty were inconvenienced by my need to take principled action.
Mustang
---------------------------------
(back to Dove - All the stuff above is from the transforma list)
If this is not a satisfactory answer, or you have further questions Lars, and actually feel a need to persue this, please do it on the Transforma list where you may get an answer, cause from what I can tell only dribbles of what's really going on ends up on tribe, and we all look like a bunch of chickens running around with our heads cut off with no idea that we're on the absolutely wrong venu to be discussin this.
-----------------------------
- Issue a statment on how tickets were decided this year, including how randomness was determined, AND how the remaining tickets will be alloted.
I have absolutely no comment on any of the other things in your email, neither for nor against, but I would like to point out that I think this point has already been answered.
On March 19th, WordPlay wrote:
---
While not comfortable with supplying the computer code on a public venue, the randomization process was based on a 20 digit alpha-numeric sequence that was provided by a random officemate, with no connection to Transformus, of one of our BOD members. This number was then applied by computer to the list of requests, which again were identified by a number rather than a name, until 1350 invites were fulfilled. No consideration was given to whether one or two invites were requested.
---
I have done some computer coding myself, so I can add some information here. Pretty much all computer programming languages have a function that is capable of spitting out a stream of pseudo-random numbers. These functions are used in all sorts of cases where a program needs to emulate a random event, like rolling a die or picking numbers out of a hat or what have you.
The "random" function can be used to generate a pseudo-random number between 1 and an arbitrary maximum. So if you had 2,000 ticket requests and you needed to randomly choose 1,350 of them, you would assign each of the requests a number between 1 and 2,000, then you would have the "random" function spit out 1,350 numbers between 1 and 2,000. Each of the numbers that came up would be a ticket request that got fulfilled.
The "random" function is called pseudo-random, not random, and this sometimes leads lay-people to conclude that it is in some way inferior to a random function, but it's really not--not for most intents and purposes. In order to generate TRUE randomness, you need an external source of information. For example, a certain company makes money by generating TRULY random streams of digits, using the background microwave radiation of the sky as a source. Most of us don't need to go to that much trouble for little things like simulating die rolls or picking names from a hat.
Pseudo-random number generation algorithms generate streams of digits that are highly mathematically correlated to actually random events. They are a very close approximation of a truly random stream of digits. A lot of research has gone into this field, because random numbers are important in areas like cryptography, not just rolling dice and picking names from a hat. If a random number generation algorithm is used for cryptography, and it turns out not to actually be random, it compromises the encryption, which is a Very Bad Thing, so the random-number-generation algorithms in popular computer programming languages are well-vetted. Every so often, a popular random-number-generation algorithm is found to be flawed, but really I think it's quite unlikely that this affected the Transformus ticket process.
One difference between a truly random event and a pseudo-random event is that the truly random event is not predictably repeatable, whereas a pseudo-random event is. The way this works is that the "random" function takes a number as a "seed". That number is used to mathematically generate the first digit of the pseudo-random number stream, and then the first digit is fed back in as the new "seed" to generate the second digit, and so on. You can therefore see that, if you enter the same "seed," you will get the same stream out each time. The stream will have the same statistical distribution as a randomly-generated stream, but it's actually totally repeatable.
It is true that the size of the seed limits the number of possible random streams, so the smaller the seed, the smaller the number of unique number streams that can exist. The seed, according to WordPlay, was a 20-digit alphanumeric string. I'm guessing it was hexadecimal (base-16) since he said alphanumeric and not numeric. This means that there were 18,446,744,073,709,551,615 possible unique streams of digits, which is plenty, I'm thinking, especially given that Transformus only pulled the first 1,350 numbers off the front of the stream. (I got that number by converting FFFFFFFFFFFFFFFFFFFF hex into decimal.)
Could the repeatability of the pseudo-random system be used to cheat the system? Sure. If you knew ahead of time what seed was going to be used, you would know exactly which 1,350 numbers were going to come up, and you could make sure that the people you wanted to get invites were assigned those numbers. This is why WordPlay points out that the seed was selected by a neutral third party and nobody associated with Transformus knew what it was going to be ahead of time.
This system is also inherently auditable. The organizers could, if they wanted to, show the database of ticket requests, including which number was assigned to each ticket request, then they could reproduce the generation of pseudo-random numbers, using the same seed that was originally used. It could then be confirmed that the ticket requests that were granted correspond to the numbers that the pseudo-random function generated. It would be very difficult to fake this out, as long as the "random" function was intact. And if the "random" function was not intact--that is, if it had been manipulated to give ticket requests to certain people and not to others--then this would be easy for a mathematician to detect.
Try Me!
--------------------------------
The process I was charged with devising and executing to fairly turn away 844 requested invites is provably fair and could be audited by someone qualified to evaluate it in good faith. I never understood the "20-digit alphanumeric sequence" gibberish. I asked Debra to ask someone unrelated to Transformus to give her a large, positive whole number, to send it to her via e-mail, which Debra then sent to me. The point of doing that was to provide a seed value for the MySQL RAND() function so that the SQL SELECT statement would have results that were reproducible, and yet the exact ordering of the outcome was arms-length from me, so that no one could say that I sought a seed value that happened to produce results I preferred. I actually cared about fairness, auditability and transparency to that degree. (About me briefly: I received honors in computer science from Oberlin College in 1987, invented a software product and started a software company in 1996 and sold part of it in 2002, thereby allowing me the free time to devote hundreds of hours of volunteer work, for no compensation, trying to help this little regional burn be what it advertised itself to be.) Turns out, my gifts and efforts at fairness were being perversely undermined by very ill-intentioned people, forcing me to withdraw my work so that they would not be used as cover for deeply dishonest activity that was going to turn the burn into a sick joke of a frat party masquerading as something that lived up to the stated principles.
Sorry that those of you who are unrelated to this dishonesty were inconvenienced by my need to take principled action.
Mustang
---------------------------------
(back to Dove - All the stuff above is from the transforma list)
If this is not a satisfactory answer, or you have further questions Lars, and actually feel a need to persue this, please do it on the Transforma list where you may get an answer, cause from what I can tell only dribbles of what's really going on ends up on tribe, and we all look like a bunch of chickens running around with our heads cut off with no idea that we're on the absolutely wrong venu to be discussin this.
posted by:
|
Rochester
|
-
Re: TO LARS
Mon, April 14, 2008 - 2:05 PM"This number was then applied by computer to the list of requests, which again were identified by a number rather than a name, until 1350 invites were fulfilled."
I don't need the actual code. Thats what I wanted. Although mustang did reveal enough information to find the exact code used.
"No consideration was given to whether one or two invites were requested. "
Actually, that "lack of consideration" leads to a constraint on the function and a change in distribution as the process moves through the list, and ultimately a non stochastic distribution. Its "bad math"(tm)
Thats exactly what I suspected, and its NOT a random distribution. The odds of a 2 ticket request being approved are higher than that of a 1 ticket request being approved. The odds change in favor of "2 ticket" requests as the total approaches 1350. The first one chosen was random, the last one chosen had the most skewed odds because of how many 1s and 2s had been removed from the pool so far. 2's simply accelerate the process towards 1350 by taking up twice as much space from the available pool as a 1.
Think of it this way. Lets say we're all 2 ticket requests. Fair right? We randomly pick some and we get to 1350 after 650 requests are pulled out of our random hat. Now lets say we all have 1 ticket requests. We have to pull twice as many (1350) out of the hat. Still fair. Now MIX The two. Lets say there are 3 invites left, the process is almost finished (1347 invites already approved). The next request picked could be a 1 or a 2. If its a 1, then there are 2 tickets left in the pool. If its a 2 then there is 1 ticket left. So when a 2 comes up, in this case it cuts the available tickets in HALF. The same goes if there are 50 tickets left. You either get 49 left or 48. Obviously we all have better odds if there are more tickets available in the pool. You can see how this process accelerates over time.
You might say... well... the 1s and 2s all look the same to the machine. That may be true, but only BEFORE they're approved. After they're approved they change the size of the remaining available pool. So if a 2 comes up, it lowers everyones odds but not its own! The 2 automatically has better odds than anybody who follows it. As do all the 2's no matter where they are in the series. So the 2 lives with some fixed odds, and the next guy gets worse odds. If the next guy is a 1 nothing happens since we have a 1:1 ratio of requests to "turns" in the random hat picking process (assuming everyone submitted one request). So a 1 doesn't change the odds because we subtract a 1 from both the available pool and from the total requests. Its balanced. If the next guy is a 2 the odds go down again for everyone who follows, but the 2 has the same odds as that last 1, while only removing 1 from the requests, and 2 from the available pool.
This bears out in the numbers, a 2 ticket request had a better chance of being approved than a 1 in the final results.
Having a human being consciously pick a seed is not a good idea. I prefer beta decay. However the seed is large enough for this purpose and I don't believe there was a prediction attack. Its possible but implausible. Anyone with the access to do it could just fake results instead. And people who dont have access can just stuff the ballot box.
"This system is also inherently auditable. The organizers could, if they wanted to, show the database of ticket requests, including which number was assigned to each ticket request, then they could reproduce the generation of pseudo-random numbers, using the same seed that was originally used. It could then be confirmed that the ticket requests that were granted correspond to the numbers that the pseudo-random function generated. It would be very difficult to fake this out, as long as the "random" function was intact. And if the "random" function was not intact--that is, if it had been manipulated to give ticket requests to certain people and not to others--then this would be easy for a mathematician to detect. "
Yes, its reproducible. Yet whats the point in reproducing it to prove it was on the up-and-up if you could have just shown us the first time? I don't see any evidence here. Just a "we're telling you to trust that we could show you if we felt like it, and trust us that if we did feel like it, it would be the way we say it would be."
The thing is, doing it the way you say you did (which I have no reason not to believe), and showing it was "random" are TOTALLY different things. So all of that is really pointless.
Yes, it would be easy for a mathematician to detect. Just so happens I can do just that. There is no need at all for reproducing the process. I just need the number series of approved/not approved requests so I can run a variety of functions for statistical analysis of probability distribution. Like I specifically asked for. You answered question 1. You did not answer question 2 or 3.
You could fake a real random series and send it to us, so to verify it and still give everyone privacy, we'd each need to know our own number, so we can verify that its not just a made up list of numbers, and that our number really does reflect if we were approved or not. If enough people just checked their own number we could have quite a bit of confidence that it is the series that was used. Since just one "wrong" result means the list is fake.
Mustang's response is clear and concise. I can tell he's got a CS degree without him even saying it. Thanks mustang, you rock. I don't know what they're doing wrong, but I am glad that you aren't associating yourself with anything you find immoral or wrong. That takes a lot of guts and shows good judgment. Hope you read this as I believe you could actually have an informed conversation about the math.
Anyway... down to brass tacks. All that I've been shown so far is that your odds are different depending on if you had 1 or 2 tickets on your request. Nobody has shown me any evidence that this was at all random. Can any information given here PROVE that they didn't pick the first 1350 requests? Or only odd requests? Or only prime numbers and then all the even ones. Any function at all could have been used and we would have no idea since its completely opaque. All we got were the final result totals, which do not indicate a random distribution at all, which is in line with what I just explained.
Ever see a lottery drawing on TV? They use machines that are kept under lock and key, select the machine at the last second arbitrarily from a group of machines, and then televise the actual picking of the numbers. This shows everyone that it was random. Statistical analysis on a run of just 6 numbers is not very accurate, but on 2000+ numbers it will be "plenty" accurate. I don't need to see the balls come up, I just need the series.
I dont have the full series for run analysis, but I do have the results... And they speak volumes. They fully support my assertions above.
The stats:
Invite Stats
Transformus 2008
________requests invites
REJECTED 561 994
approved 704 1279
unallocated 1429 (I think this is the unpaid-for tickets but I dont know what it represents and its not being used in my math)
capacity 1500
1265 requests were made for 2273 people. Thats 1.797 invites per request on average for the whole bunch.
1.817 invites per request for the approved.
1.797 invites per request on average.
1.772 invites per request for the rejected.
1.817 > 1.797 > 1.772
Theory aside, the results show that your odds were better with a +1. Still don't believe me? Fine... lets go!
704 people who filled out the form got in, but 1279 people have invites. So 575 people are +1s and 129 are singles. 129 + 1150 = 1279 so 1150 people were part of a +1 request.
561 people who filled out the form were rejected containing 994 total invites rejected. So 433 +1s and 128 are singles. 128 + 866 = 994 so 866 people were part of a +1 request and rejected.
In total, there were 2016 people on +1s requests. And 257 singles requests. Of those 1150 people got in on +1s requests, and 129 got in on singles. 866 were rejected on +1s, and 128 were rejected as singles.
55.653% of people got in overall. (1265/2273)
57.044% of people on +1 requests got in. (1150/2016)
50.195% of people on single requests got in. (129/257)
We must take the sample sizes into account. Was this by chance just because the sample sizes were different? No it wasn't. This means there is a significant difference between the two percentages that is not due to chance. This can be (and just was) calculated with the z-test. en.wikipedia.org/wiki/Z_test
Taking the percentages, sample sizes, and total population into account, z=2.335
Formula: z = |p1-p2 |/s
where:
p1 = percentage 1
p2 = percentage 2
s = sqrt(p(1-p)/n1adj + p(1-p)/n2adj)
p = (p1adjn1adj+p2adjn2adj)/(n1adj+n2adj)
n1adj = (N-1)n1/(N-n1)
n2adj = (N-1)n2/(N-n2)
n1 = sample size 1
n2 = sample size 2
significance test:
z >
2.576 for 99% level of confidence
1.96 for 95% level of confidence
1.645 for 90% level of confidence
2.335 far exceed 95% confidence. These results are statistically significant. You REALLY DID have a better chance on a +1 ticket than as a single. And you would get the same difference in percentage chance every single time this "random lottery" is run (with a 95 percent confidence).
"Try Me! "
Consider yourself tried.
I trust that mustang did this as best he could because he was faced with a constraint. The problem is, you CANT do this fairly. Its not possible. So yes, you can claim its "the best we can do", but you cant claim its random pseudo-random or fair or equal. You have a significantly higher chance of getting in with a +1.
Mailservers are soooo 1992. No way am I signing up for that. If the people in charge of this bruhaha don't read this tribe, then no wonder they're out of touch with whats going on! Thank you for suffering through the mailing list and forwarding my questions.
However, my questions 2 and 3 were not answered.
2. Please give me the series so I can do statistical analysis to prove it was random (this includes any reserved tickets out of the 150 reserved for that purpose!)
3. Can you prove nobody entered more than one request. If so how? (the how is important since I'm quite certain you cant) -
-
Re: TO LARS
Mon, April 14, 2008 - 7:04 PMLars,
You are missing the point.
The point is - THIS DISCUSSION IS NOT TAKING PLACE ON TRIBE! I cut and pasted this information from the transforma list to tribe to show you the responses to the "random" question on that list. You will NOT get your answers here. If you want to persue this further, you need to take it to the transforma list, because the people taking place in this discussion are not posting here.
In addition, the BOD has stated they will not be doing it this way in the future. Furthermore there is enough shit being stirred up over things like possible fraud, hording tickets in PAST YEARS (btw did you realize that was the big scandal here? not that THIS year was unfair, but that PREVIOUS years were? that people are raising such a fuss because the people who "should" have gotten in didn't...and then people were asking wait, why did no one complain in past years? Maybe its because tickets were being reserved for friends etc. Again, this stuff is all on display on the transforma list), other accusations...the issues you keep raising? People have moved way past that.
It's not going to change the fact you didn't get a ticket. You've just got to come to terms with it and hope you get one later from someone who isnt' going. Your point has been taken, the BOD knows a lot of people were unhappy, they're changing it for next year. So please...either post your concerns to the transforma list, where you may actually get a response, or get over it. But you aren't going to get your desired outcome of an answer here. -
-
Re: TO LARS
Mon, April 14, 2008 - 8:33 PM"THIS DISCUSSION IS NOT TAKING PLACE ON TRIBE!"
I think you mean THAT discussion. This one is quite clearly here. :) See... look at the top of the screen, it says "tribe"
Seriously, its ok. I understand that they're discussing everything over there. This has been stated over and over by many people. I don't know if you just want me to shut up, or if you think its actually useful for me to go over there.... but I am not going to.
I do appreciate you forwarding the information here, and I don't expect you to be my monkey and continue to do it. If thats what you mean. I really hope that people don't want this discussion suppressed here simply so fewer people are aware of how truly fucked up the state of affairs are. I don't think thats your intention, and I certainly hope it isn't.
They cant do math AND they cant read the tribe for their own event? The way they're presenting themselves... I don't think I want to join their mailing list anyway. Hell, the semi-private mailing list is even more elitism. If they cant say it in public, then something is wrong. If people tell me I cant say something about it in public, then something is wrong.
Yes, I am very aware of the problems in the past. They handed out tickets at work/play like candy. The BOD reserved tickets, etc. I know. I think that was unfair but still a hell of a lot more fair than what we have now. I know about it... whatever.
I'm not really concerned with "people didnt get a ticket" complaints from prior years because that has nothing to do with this stupid, rigged, not-at-all-random lottery. Those complaints are no different than they were before or than they are for other events. Thats just your standard "noise". This stuff is just plain WEIRD and totally unique to this burn this year. The complaints of favoritism and hoarding are valid, but I'm picking my battles. I don't see how this is linear and people have "moved on". I don't think anyone has addressed this issue at all. You're not, you're just dismissing it as being unacceptable for tribe. As wordplay said, "try me". He seems pretty confident the math is right... which is a major assumption that just happens to be false. Nobody but me has even questioned it. And since nobody answered it, I put a great deal of intellectual effort into answering it MYSELF. So there you go. There ARE answers. This is useful.
Why does everyone have this knee jerk reaction "oh you didn't get a ticket so you're mad so therefore what you're saying isn't true or doesn't matter" I still don't get that.
Is my math wrong because I didn't get a ticket? If so, please correct it.
I didn't have a ticket at this point last year and I wasn't complaining. Show me a post where I complained last year. Go ahead. I'll wait.
The only complaint I made this year about ME not having a ticket is that I volunteer, and with fewer volunteers the cap must be lowered. Thats all. I was done with that 3 weeks ago. Why does everyone read my comments and see "blah blah blah I'm a whiny baby who didn't get his rattle, blah blah blah"??? This is beyond me. Can people not read? Or do they just hallucinating a bit too early?
I know there is a chance I'll get one, and that the real number of people who want to go is way lower and that many people cheated thus inflating the numbers. I'm tenacious and I'm not writing off the event. So its not a matter of "getting over it".
I'm pissed that up to 800 people didn't get in and more specifically WHY. I don't know why everyone thinks I'm so self-centered. I am pissed that the people who did get in have survivors guilt. I'm pissed that the organization lied to us intentionally in an attempt to prevent us from cheating the lottery, I'm pissed that they have no method for determining if people cheated by entering more than one request. The fact that I didn't get accepted is arbitrary and tangential to all of that. LOTS of people didn't get in.
I don't like all the people who obviously have invites who say "oh, there is nothing we can do, we just have to make the best of it", uhhh... yeah actually there are things we can do, like figure out what the fuck happened and fix it. I dont see how me having a ticket or not changes how fucked up this situation is or its fix-ability.
So if people who got a ticket make an argument then its taken seriously, but if someone who didn't get a ticket says something its automatically dismissed? Then someone give me a fucking ticket, then maybe people will start listening to what I say instead of dismissing it. Because I damn sure am going to keep saying the same thing even after I have my ticket in hand. Or do I just have to lie and say someone gave me a ticket for you to start taking me seriously? Weird, being dishonest would make me more credible... Whoda thunk it?
"the BOD knows a lot of people were unhappy, they're changing it for next year. So please...either post your concerns to the transforma list, where you may actually get a response, or get over it. But you aren't going to get your desired outcome of an answer here."
I've gotten lots of answers here already. You cant predict the future. Although, I could be wrong, in which case please give me the lottery numbers for a few states next week. :) I'd like to win 3 states on the same day, that would rule.
I'm not trying to make the point that people are unhappy, thats blatantly obvious to anyone. I'm trying to make the point that there is a FATAL FLAW IN THEIR MATH. Has anyone else made this point? NO. So if I don't who will? Apparently nobody.
I know they're not doing this next year, because people would cheat even more than they did this year. Thats why they lied to us this year to attempt to prevent that. I don't need to read transforma to figure that much out. The question is, why fix it next year if we know its a problem now? Allow people to get away with cheating the system? Leave this stupid lottery thing that upsets people? Spend a MONTH writing code to salvage something with no value at all (a random list of numbers)? Why? Why not just instate the sales starting next week and just do it like they always did? Fixing it now is totally possible. Their invites aren't binding. Did anybody sign a contract? NO. The few people who paid can get a refund and they'd have no legal recourse. So why wait until next year? Is there a stone this is set in that I don't know about? They could have avoided this train wreck long ago and they still can, yet they're still slogging forward neck deep in the big muddy. Its kind of like iraq. :)
So unless someone over on transforma is already pointing out the flaw in their math, then I suggest you tell them that this discussion is happening HERE and that they need to come over here if they want to hear about it. And if they don't want to hear about it, me talking about it there wont do any good now will it?
I don't care if the people "in charge" hear it or not. I just want it publicly known that the math is fatally flawed, and to demonstrate actual evidence of that fact. I have done so. Putting it in a closed mailing list doesn't help me with that goal. I am more concerned that the public read it than the actual decision makers, because its clear they don't really want to hear what we have to say, or they'd be here. -
-
Re: TO LARS
Mon, April 14, 2008 - 8:50 PMI think Dove's point is that you're making your point in the wrong forum. Valid points, yes, but the wrong venue. -
-
Re: TO LARS
Mon, April 14, 2008 - 9:22 PMIf you give me permission to pick up what you've said and stick it on that list, I'll do it.
Don't mind being a monkey sometimes.
Dang Lars you are SMART. All that math is WAAAAY beyond me. Duuuuh I can draw pretty pitchers tho. Gnarf.
My husband had noticed the math problem as well, but I couldn't understand what he was saying either, just that it sounded like what you said (which in my mind is translated something-something-something about how every application with two approved makes those with one less likely to get it, something something something).
But, we're certain that it was a math mistake and that Mustang believed he had come up with a fair system. I also think that your being able to analyze it this way is an excellent example of why things should be handled in a more open way...someone may be smart as hell (like Mustang), and totally well intentioned (like Mustang), but none of us are perfect, any of us can make mistakes or not notice a glitch, and more people to look at any given problem can only lead to a better solution in the long run.
XXXOOO,
Me
-
-
Re: TO LARS
Mon, April 14, 2008 - 11:05 PMI'm a drawer/painter/photographer too. Its cool having two brains. Although I have to say, your drawings are more appealing than mine :)
You can post that anywhere you want. If they want to discuss it behind closed doors they're free to do so. Or ignore it behind closed doors... whatever they want to do with their radical expression. :)
I strongly suspect at least some of them are reading this though. (how could they have posted here if they aren't reading?) I am sure they're trying to keep the discussion off of this public forum. I understand... I would too if something I was responsible for planning blew up into a 300+ comment thread about how fucked up what I had done was. Totally empathize there. I'd be very quiet here too if I were in their situation.
I realize that everyone cant understand all the math. (everyone has their limitations on what math they understand, mathematicians included) I tried to explain it in logical english instead of mathematical notation first to give it a slightly broader audience, but I do realize that most people can only look at my results at the end and say "well, shit, I cant dispute that cause I don't understand it", and I see no way around that really. I tried my best. Just trust that this is a big group, and someone here has the knowledge to understand it and possibly dispute it. Which I hope they do if they can, I'm more interested in the truth than fooling myself into thinking I'm right when I'm actually not.
Feel free to forward it to any mathematicians or statisticians that you know. Probability theory is complex, and more eyeballs are always good, like you said.
The statistical analysis is only there so people cant say "bah! sample size! not significant!". I'm covering all my bases and not all of the post is appropriate for all people. Its hard to be clear AND complete.
Mustang came up with the fairest system he could with the constraints they gave him. If he is aware how those constraints affect the results I don't know, but there is no way he could have made a random system any better if the +1 ticket rule must be included AND he must create a random selection process. I do not fault him. His response was clear and understandable and I believe everything he said is honest. Nobody is infallible and even if he DID know what the results would be, there really wasn't anything he could do about it given his circumstances.
I fault people who insist that it was "random" and "fair" without testing it or understanding it. I admit, the deviation from a random distribution is small, but it is real and can be proven to be significant and not some random occurrence. It may not seem like a big deal, but it sure is a big deal to the 5 or 50 people who were rejected simply because they didn't and an extra +1 that they didn't need. I'm PROBABLY not one of those people, but they exist undoubtedly. I'll go ahead and assume I am just a victim of bad luck, and not a difference between my odds and someone who added a +1.
I want to make it clear, I'm just explaining what happened. I know that nobody wants to hear someone comlain about something without providing a better solution. If they insist on +1s as well as a random system, then this is the best they can do and there is no way I can improve upon what mustang did. (which is why I was able to guess at the exact method he used without asking, I only asked to make SURE I had it right, assumptions aren't allowed in math)
I know that in a do-ocracy complaining without providing solutions is frowned upon. However, I do want to point out the ACTUAL results of that system more than I want to complain. And the only suggestions I can give are to do away with +1s and keep the random system (which I'm not too fond of but it is more equal), and of course my favorite suggestion "if it aint broke dont fix it"... i.e. go back to the old method of first come first serve ticket sales. Which is exactly what they plan to do next year AFAIK.
I say that even though I didn't get a ticket in last year's sales. Note that I didn't complain at all. I made ONE comment before the event last year, and that was to get my last minute ticket literally days before the event. Right up until the end I didn't complain or even beg!. Just go look at the last dozen or so posts before the event last year and you'll find my single comment and nothing else, no matter how deep you dig. I got "screwed" by the system last time, but it was a fair system, so I had nothing to complain about.
Oh, and I want you to understand this. I'll give a simple explanation. Its not directly applicable to this, but it gives you an idea of how you can be confused by odds.
You remember the game show lets make a deal? You'd pick door 1 2 or 3. Lets say you pick door 2. Then Monty opens up door 3 to reveal a goat. You're then given the option of sticking with door number 2, or switching to door number 1. One has a goat behind it, the other a new car. It seems logical that you don't change your odds if you change your guess.. but in fact you DO have better odds if you switch your guess! It doesn't make sense, but the math supports it. Basically, in your first choice, you had a 1 in 3 chance. Once he revealed more information (the goat behind 3), you now have a 1 in 2 chance, which is better. But you only have that chance if you switch your choice. Otherwise you're still the way you were before, with a 1 in 3.
If you don't believe me, go play the game over and over. You'll figure it out by learning. :)
www.stat.sc.edu/~west/java...eaDeal.html
It also includes a better explanation than I just gave. The way that this equates to what's happening here is that every time a 1 or a 2 ticket is chosen, it changes the odds as we narrow down how many doors are left. Not that we have a choice in the matter. Just that the odds change as information is revealed. Even if its only revealed inside the system that we cant see. Its still there and still real.
Anyway, thanks for trying to understand it and actually adressing what I was talking about, instead of if I should be talking about it or where I should be talking about it :) Its refreshing. -
-
Earth to Lars
Wed, May 7, 2008 - 8:10 AMI'm not going to take the time to refute or confirm everything Lars has to say ... I'm of the impression his work should be taken summarily without thought toward every detail. He clearly enjoys typing and anlayzing. If you enjoy reading his posts, you might consider reading Johnathon Edwards' Sinners in the Hands of and Angry God and DesCartes' Meditations on First Philosophy (literary source of "the Cogito"). But I don't enjoy reading such things in detail, although with regard to these works by Edwards and DesCartes, I have read them in detail and have come to summary conclusions about their values.
Lars says:
You remember the game show lets make a deal? You'd pick door 1 2 or 3. Lets say you pick door 2. Then Monty opens up door 3 to reveal a goat. You're then given the option of sticking with door number 2, or switching to door number 1. One has a goat behind it, the other a new car. It seems logical that you don't change your odds if you change your guess.. but in fact you DO have better odds if you switch your guess! It doesn't make sense, but the math supports it. Basically, in your first choice, you had a 1 in 3 chance. Once he revealed more information (the goat behind 3), you now have a 1 in 2 chance, which is better. But you only have that chance if you switch your choice. Otherwise you're still the way you were before, with a 1 in 3.
Lars is wrong.
If Lars or anyone else would like to discuss real chance principles and combinatorics with me at Alchemy then look for the flaming drum and hope to find me at a sober moment. Or, if I can find a ticket, I'll be at Transformus. In the meantime, I recommend reading some Feynman.
BMac -
-
Re: Earth to BMac
Wed, May 7, 2008 - 9:18 AMDamn ... I think Lars is right ... still working on it ... I don't believe the applet programs ... can't review the code ... working it out now in real space with coins ... damn it ... I hate it when I'm wrong saying somebody else is wrong ... aaahhh! ... ok, if Lars is right, do I get to keep my flaming drum? ... I'm just a fucking clown! ... whaada I know? eh? ... aaahhh! ... i'm on fire! ... roll me around, put me out! ... now a little kiss ... some tongue ... yes, better now ... sure you can invite your friend, it's all good baby ... -
-
This is the maximum depth. Additional responses will not be threaded.
Re: BMac to BMac
Wed, May 7, 2008 - 9:34 AMOk ... I'm back ... and Lars is wrong.
However, I've only got him on a narrow technicality based more on his wording in presenting the explanation of the apparently infamous "Game Show Dilemna," with which I was previously unfamiliar. Lars is generally correct that the switch strategy prevails.
And yeah, I've got Edwards and DesCartes on technicalities as well.
So let's get back to the weenie roast ... where's my kiss? -
-
Re: BMac to BMac
Wed, May 7, 2008 - 3:48 PMActually, Lars is right that the right strategy is to switch, but his statement that your odds if you switch are 1/2. The real odds if you switch are 2/3.
Let's examine the case where there are 100 doors. Your chance in picking the right door is 1/100. If I then gave you the option to switch from the door you picked or every other door, it's obviously 99/100 and you would switch in instantly.
The opening of the doors is a distraction. The operator of the game show know where the goats are and where they aren't. When they throw open the 98 doors, they are either picking the 98 doors with goats leaving you with the car behind the door, or you picked the car in the first place and they throw open 98 arbitrary doors (as all of them have goats). If you stay, you still have 1/100 chance to win. If you switch, you are selecting the 99 other possibilities presented as a single door, giving you a 99/100 chance. -
-
Re: BMac to BMac
Wed, May 7, 2008 - 5:55 PMYou guys are killin' me with this thread...lol! Pretty fucking interesting in a way that makes my head spin a little..
-
Re: BMac to BMac
Thu, May 8, 2008 - 5:06 AMThe monty hall problem is dependent on the fact that the host knows which door has the car, so he will always reveal a goat, and past outcomes (the previous choice) effects the host's behavior. If the host randomly opens a door with no knowledge of what is behind it, the odds are the same whether you switch or don't switch.
The 1/2 vs 2/3 thing may be an matter of confusion in Lars's statement, I don't feel like going back to reread it to check.
Your chance of choosing correctly in the second round of slection is 1/2 when it is isolated from the first selection round. However, the overall probability after both rounds of selection is 2/3 if you switch and the host always reveals a goat.
-
-
Re: BMac to Earth
Thu, May 8, 2008 - 7:35 AMY'all might be missing the point of my post.
Lars has been banging other people around over math skills while his treatment of a classic challenge is wrought with gibberish.
Nonetheless, I thank him for bringing the challenge forward for my consideration and he gets partial credit at least for knowing the switch strategy prevails. That's all that really matters in the game show scenario ... right?
BMac to Bink:
It doesn't matter whether the host knew which door has the prize once it is established that the host has opened a door behind which no prize exists. Lars presented us a one-shot scenario. To disentangle this challenge, I set aside the on-line simulators and website chatter from various math departments and played out some iterations in real space. You might note that I recommended reading of Feynman.
That's all the care I have for counting the angels that dance on the head of a pin ... y'all debate this all ya want ... but I'm down to this:
Lars is wrong.
Lars, check your own math before you come back here and grind on my tribe ... cause the BMac is in the house ... and I eat pie!
BMac -
-
Re: BMac to Earth
Thu, May 8, 2008 - 6:05 PMJust saying I'm wrong doesn't make it so.
You have to show why I am wrong. I'm not even sure what you find fault with in general.
I'm quite familiar with Feynman. Although he was a physicist and this is statistics.
-
-
-
Re: BMac to BMac
Thu, May 8, 2008 - 6:15 PM -
-
Re: BMac to BMac
Thu, May 8, 2008 - 9:18 PMLars, from your link:
<i>in the usual interpretation of the problem the player should switch—doing so doubles the probability of winning the car from 1/3 to 2/3</i>
not 1/2.
Bmac:
>It doesn't matter whether the host knew which door has the prize once it is established that the host has opened a door behind which no prize exists.
Actually it does matter that the host knows which door has a prize in it. If the host has the ability to throw open a door with the car behind it, then switching does not have the effect of choosing the other two doors. Either he reveals a car and you have no chance to win, or you end up with even probability (1/2) if you switch or don't switch. There's no strategy involved.
The point is that switching is the same as selecting the other two doors because the host will always open a door with a goat. Leaving you with your 1/3 chance of getting it right in the first place, or 2/3 chance the car is behind the door the host purposely didn't pick.
-
-
Re: BMac to BMac
Mon, May 12, 2008 - 5:15 PMI did not mean to imply that overall, adding the odds you will not have 2/3 odds. I meant in that particular instance, you've gone from a 1 in 3 chance, after information is revealed to having a 1 in 2 chance if you switch. Your first decision was made 1 in 3 and will always be that way, the only way to change your odds is to switch AFTER information is revealed.
If you play the game forever (always switching) you'll win 2/3 of the time. If you play once and switch, its 1/2 odds because you have one choice and 2 doors. If you play once and don't switch, its 1/3 since you ignored the goat information. Sure, they're not the "real" odds because of whats going on behind the doors and what the host does. The "real" odds are calculated by adding up all of the possible odds. The "real" odds are not as important here as what CHANGES about them. All you're saying is that your odds increase by more than what I claimed. Point is odds go up. I just didn't feel like explaining exactly what the sum total odds are since its irrelevant.
The way that my point relates to the ticket lottery has been lost here. We're getting hung up on what the host does and why, which is not part of my analogy. The point of this is, if information is revealed, it changes the odds for any decisions made later on that take that information into account. Even if they are truly random choices.
I'll try to make this as short and simple as possible. If you think my math is wrong. Show me another mathematical explanation that will account for the statistical deviation. So far the arguments I've gotten have been the equivalent of something not-unlike the grammar nazi's claiming I made a typo and thus my main point is false or lacks importance.
As fun as this is, I don't intend to explain math to anybody ever again. So far I've been dismissed with "thats not important because its long and makes my brain hurt". Thats why I tried to include this "everyday puzzle" as an explanation of how odds change as information is revealed. Unfortunately it just serves as a distraction.
I may have done better to simulate the lottery exactly. Showing the current odds of 1's and 2's as it iterates through a random set and its total of 1's and 2's summed is 1350. You can graph the odds for each population as they change. I considered it but people could make claims about the process I used for the simulations. The final statistics of the lottery are indisputable and an analysis of them I felt would be more convincing and is totally transparent. You can sit down and calculate this out yourself and compare your results to mine. At this point I'm not coding something in an attempt to convince people of what already believe and dismiss as irrelevant, or that they don't even want to believe in the first place. Especially pointless since this lottery will not be used next year (yet somehow its acceptable to be used in the second round?)
It seems to me that the engineer who designed this system knows of my mathematical analysis refuting their claims. Yet I see no defense. This means there isn't one yet, and given the time that has passed there probably never will be one. So if you don't understand it, accept it tentatively until refuted by those who state that it is in fact fair and equitable and can demonstrate their assertion through math. Someone designed this system right? Let them stand up and defend it. What I notice when I see people talking about my actual math is that its correct but that they don't care that the "random" system actually has a statistically significant bias.
What people see as me insulting the mathematical ability of this group is not meant as an insult... merely a lamentation on the expected (and apparently rightly so) lack of mathematically rigorous testing of my work here. I can't expect everyone to be a mathematician. I was hoping we might have one or two though. -
-
The Lars Challenge
Wed, May 14, 2008 - 7:20 AMLars says:
As fun as this is, I don't intend to explain math to anybody ever again.
BMac says:
Please hold to your intention ... after all ... Lars is wrong.
Lars says:
I did not mean to imply that overall, adding the odds you will not have 2/3 odds. I meant in that particular instance, you've gone from a 1 in 3 chance, after information is revealed to having a 1 in 2 chance if you switch. Your first decision was made 1 in 3 and will always be that way, the only way to change your odds is to switch AFTER information is revealed.
BMac says:
That's just gibberish. You haven't tried the modelling I recommended. You haven't comprehended the websites you recommended. The point of my physical modelling was not to calculate probabilities or run many iterations. The point of my modelling was to better understand the Lars challenge, which is a single iteration game. This very Feynman-esque point resides in the divide between math and physics. For example, Lars says "If you play the game forever" ... but I'm a physicist ... so I know I can't play the game forever. I also know that a thousand monkeys typing for a thousand years won't produce a single Shakespearean play ... but go ahead and "calculate" the probabilities if you must ... just please don't share your results ... after all, you promised you wouldn't, and don't forget ... monkeys don't actually live for a thousand years ... do they?
Did you know that the BOD had no intention of playing any game forever when they tried to establish a fair ticket distribution system?
Cheers,
BMac
-
-
-
-
-
-
-
-
-
-
Re: TO LARS
Mon, April 14, 2008 - 9:37 PMYes I know.
She thinks I want to convince some people "in charge" or "in the know" to give me some information or make some changes. That would be nice, but thats not what I'm trying to do right now.
I'm ACTUALLY trying to show EVERYONE who applied for an invite what their ACTUAL chances were. Posting it in transforma defeats that purpose because most of you wouldn't see it.
Secondary to that I also want to make sure that what I'm saying is falsifiable by people who can read and comprehend the math, If someone is in a position to dispute my math, the chances of that are greater in a larger group than in a smaller one. I'm guessing we have more mathematicians here than on the mailing list.
You can see from wordplay and mustang's response that they assumed my question was inferring some kind of malicious activity and not just plain old run of the mill incompetence. Which we all know is much more common. The reason they think its malice is because thats whats being accused back and forth over there, and I'm really not interested in that discussion. This is different, and they're seeing it through the eyes of their discussion. I'm not really interested in that discussion or having my points conflated with it.
Its no big deal. Like I said, this is for everyone to know what REALLY happened in the lottery. I would have posted this sooner but I wanted verification that the actual algorithm was exactly what I suspected it was before I could be sure about what I was saying.
Anyway, this is directed at you, and everybody else. Note how it says "your chance was X percent" in the post. I'm talking to everyone, not just the team leads or the BOD or active discuss-ors :) Everyone who requested an invite deserves to know this information.
Seriously, who volunteers to run an event like this and doesn't read the tribe for it? Thats absurd. They're reading, they're just not responding. :) Well except when they do respond of course :)
They want to keep everything behind closed doors. I don't. Pretty simple eh? -
-
Re: TO LARS
Thu, April 17, 2008 - 9:48 AM"Seriously, who volunteers to run an event like this and doesn't read the tribe for it?"
From my perspective, the more web sites I have to go to in order to keep up with communication, the harder it is for me, which is why I prefer mailing lists, which Gmail nicely presents to me all in one interface. As forum software goes, Tribe really bugs the shit out of me. No way to track threads. No way to quote text or use HTML to mark up text (italics, bold, underline).
-
-
Re: TO LARS
Tue, April 15, 2008 - 6:46 AMThe thing is. . . the discussion is happening here too.
Folks can keep pointing people back to other forums and venues. . .
But there clearly is a discussion happening here.
And if people (BOD) want to get information out, defend actions, clear things up. . . etc. -- they'd be well advised to respond in any and all venues. . .
You have to go were the people are talking. . . -
-
Re: TO LARS
Tue, April 15, 2008 - 7:38 AMYes, A discussion is happening here.
No, the "official" Transformus list is not here and thus the people who are responsible for many of our complaints are not bringing the discussion here.
We just had this same arguement on the PDF tribe too. A lot of folks on the burningass list (the PDF planning list) didn't realize that so much discussion was going on on tribe. Some didn't even know tribe was used as a discussion forum in the first place.
Lars, my statements to you were in response to the fact you were asking QUESTIONS (and also making accusations, to which people could not respond; you should also re read the things you wrote like yelling FRAUD on other discussions before claiming that you weren't saying things that would make folks defensive - I would get defensive too) that no one on here could answer. I feel they can be answered on the transforma list, hence why I said you should take it there.
I would also consider who and what you are calling "elitist" when you make statments like "mail servers are sooooo 1992." That sounds pretty elitist to me, especially since a lot of people can't access tribe at work and prefer mailing lists which they can receive. Transforma is not a private list, you just have to sign up and you're in. Kind of like tribe, no? -
-
Re: TO LARS
Wed, April 16, 2008 - 12:53 PM"to which people could not respond"
To which people choose not to respond.
"you should also re read the things you wrote like yelling FRAUD on other discussions before claiming that you weren't saying things that would make folks defensive "
I've read everything I wrote. I wrote it too.
Do you knot know who I was accusing of fraud? People who entered more than one request. Are you defending them? Is it bad if cheaters are defensive?
The question is, did YOU read what I wrote?
"I would also consider who and what you are calling "elitist" when you make statments like "mail servers are sooooo 1992." That sounds pretty elitist to me, especially since a lot of people can't access tribe at work and prefer mailing lists which they can receive."
Mail servers ARE so 1992. How is that elitist? Am I elitist because I have electricity in my house and oil lamps are so 1800s?
Anybody who cant access tribe at work simply doesn't know what a proxy server is. They should be doing their job and not messing around with private email if the company has such policies anyway. I don't see your point. Who EXACTLY only has internet access at work and not at home who should be reading this?
"Transforma is not a private list, you just have to sign up and you're in. Kind of like tribe, no?"
Tribe is open and can be read by the public without signing up. Transforma is private and you have to sign up. If you're implying that they're alike in that you have to sign up for both you're wrong. Look:
Your profile:
http://64.233.169.104/search?q=cache:hHvCR9WKTy4J:people.tribe.net/dovelette+tribe+dovelette&hl=en&ct=clnk&cd=1&gl=us
This very thread:
http://64.233.169.104/search?q=cache:3Q1ofGZu0BMJ:transforma.tribe.net/thread/4e12205b-988e-43b7-b64b-ca82a5234d4a+tribe+dovelette&hl=en&ct=clnk&cd=9&gl=us
I actually called the list closed, but I also called it private, because it is private for about 6 billion people on earth right now.
I want peer review. Closed sources are not conducive to that.
-
-
-
-
-
-
