Introducing: The Amazing Anonymous Bitcoin Lottery

[nelisky]

I feel very torn about how to best move forward, and I don't want to get into the "pick 3" until I leave this one running on its own. So, my thoughts right now are that we need to make sure that prizes are interesting for new draws, rollovers exist to increase capture of new betters but guaranteed jackpots are needed to share the wealth. Also, if jackpots increase we need to make sure that it doesn't pay off to just buy all the tickets and walk away with guaranteed profit...

My only defense against that latter part is the 2% bounty (unless the bounty recipient is the one doing this, of course ) so the jackpot must always be smaller than the 2%, making sure that if you buy all the tickets you end up with a loss. This is supposed to be about luck, not good math skills!

So far, I have:

- No duplicate bets
- Rollovers of only the 1st prize, all others get paid in the draw
- 20% of rollover goes to side pot, used when jackpot is won
- Whenever the side pot reaches an interesting level (say 1000 btcs) there's a guaranteed jackpot draw, where the 1st prize is also divided.
- 5 character bets with prize distribution:
5 hits: 25%
4 hits: 10%
3 hits: 12%
2 hits: 16%
1 hit:  35%
bounty: 2%

You'll notice that I reduced the 1st prize in half and upped all others, effectively distributing 73 percent of the pot in every draw, thus making it worth it to play right from the start. And for someone to guarantee 73% of the pot they'd have to spend 65536, 1310 of it going to the bounty, and 16384 going to 1st prize slice. Assuming one does not get first prize (although chances are 1 in 16, which is not too bad) the accumulated jackpot would need to be around 20k to be worth it. Because 20% of rollover goes to side pot, that pot reaches 1000 when the jackpot reaches 5000, thus eliminating that chance (we could go higher with the side pot, but for now that would mean too much time between guaranteed jackpots, I guess).

Would this work for you? I locked the current draw as only my 100 BTCs are in so far, but will unlock if you rather have things as they are, but I really think that we need a better plan, and I like this one

[1715037968]

1715037968

[1715037968]

1715037968

[FreeMoney]

Since we're back to 1 in 1048576 chance of winning the jackpot per ticket this is essentially always going to grow until you do a guarantee. So now you'd just have some lotteries that are a little -EV and some that are a little +EV, my buying will be something like 5, 5, 5, 5, 5, 1000. Which is fine with me, but I'll probably only get excited about 6 times a year or something.

I don't know how you are going to remove math skill without just making it -EV all the time by raking it hard. Well, someone could back it and the payout could be static. That's 4096BTC with 3 digit, 1BTC tickets. The backer takes substantial risk though and would porbably have to be paid. I'm not saying you should do this, I don't think that it's important to eliminate skill, we just need to find a way to not make it a raw deal to get the ball rolling.

Maybe you are onto something though, the only issue is the fact that 5 will only get hit once a decade if we buy 1000 tickets each time. Maybe keep it 3 digits and pay out every time in the 75-80% range? Roling over the rest. There will still be incentive to cover the board and lock in profit in the amount of the jackpot.

This is actually pretty hard to design. I'll keep thinking.

[vess]

I just read the New Jersey Pick 3 lottery rules over. They have a simple system:

1) 50% of the ticket sales go to the prize pool
2) (they have a bunch of ways to win, but I'll unitize them all to a number of 'Wins')
    They calculate Ticket intake * .5 / # Wins => Prize Share
3) Each Winning bet gets a prize share.

To compare, the powerball system
A) seeds the jackpot with an amount, ($10mm), advertises twice that as the 'annuitized jackpot', and proceeds as follows:
B) 50% of the ticket sales go to prizes.
C) Roughly 58% of the prizes go to the jackpot, won maybe once a month. The other 42% are for the smaller prizes.


As noted by the powerball website, powerball is designed for people who are motivated by a  big jackpot. Presumably Pick 3 is designed for people who are motivated by a small, frequent (it runs twice a day) payout.

So, either of these seem like fine setups -- hard-headed expected-value-discounted-cashflow-type people only play powerball when the advertised annuity is up in the $250MM+ range, and everyone else plays when they like the sound of the jackpot size. No math people play Pick 3, I'm sure.

My takeaways:

I) I think your lottery needs to choose a personality: is it going to be a big prize, monthly-ish win type situation, or a frequent small-scale wins one? I don't mean monthly drawings, just likely 'big' wins at once a month.
II) You probably should have a minimum pot for jackpot winners: this ensures people want to play more often. You would fund this out of your own funds to start, and then whatever you decide your rake should be, eventually.
III) An out of the box idea I had is that you could write some sort of rules algorithm and allow anyone to run their own lottery; you'd take, say 10% off the top, and lotto designers could choose their own payouts, jackpot rollovers, etc. Natural selection will tell you the most successful lottery designs quite quickly, I imagine.



As a sample 'powerball'-esque lottery: if you had a jackpot minimum prize of 10 BTC and two drawings a week, 100 people playing per drawing, odds of winning say 1 in 1000, then you'd see the following:

Week 1, Drawing 1
20 BTC goes in: 10 covers the jackpot: 10 BTC goes to rake
80 BTC goes in: 40 goes to rake, jackpot is now 50.

Week 1, Drawing 2: Jackpot 50
100 BTC goes in: Jackpot is now 100 (no winner)

Week 2, Drawing 1: Jackpot 100
Odds of nobody winning so far: (900 / 1000 ) ^2 = 81%
More people get interested: 150 buy. Nobody wins.

Week 2, Drawing 2: Jackpot 175.
Odds of nobody winning so far: 69%

I don't know if this would be appealing to people or not. I bet if the jackpot were 1,000 BTC, odds of winning say 1 in 5,000 to start then it would be appealing. The simplest way to keep math-oriented speculators out is to keep the expected value down, except in rare  circumstances.

It's gotten too late for me to add more to the conversation, so I'll summarize a TL;DR -- Copy some successful lotteries! Use the rake to provide a guaranteed pool! (Possibly: let people use your software to host their own lotteries).

[nelisky]

Since we're back to 1 in 1048576 chance of winning the jackpot per ticket this is essentially always going to grow until you do a guarantee. So now you'd just have some lotteries that are a little -EV and some that are a little +EV, my buying will be something like 5, 5, 5, 5, 5, 1000. Which is fine with me, but I'll probably only get excited about 6 times a year or something.

Remember the side pot, or rake or whatevery. After you do the 1000, and the whole of the 4000 BTC pot gets distributed, there's a 1000 pot priming, so it's not back to square 0.

I don't know how you are going to remove math skill without just making it -EV all the time by raking it hard. Well, someone could back it and the payout could be static. That's 4096BTC with 3 digit, 1BTC tickets. The backer takes substantial risk though and would porbably have to be paid. I'm not saying you should do this, I don't think that it's important to eliminate skill, we just need to find a way to not make it a raw deal to get the ball rolling.

I know the math skill is always there, and the math simple enough for those so inclined, so my only defense is to make it too expensive to have a guarantee of profit. This is fun for me, and if I had 10k coins I'd certainly be able to bootstrap this is a much more appealing way, but I don't. And I'm not asking anyone to do so when there is nothing to give in return

Maybe you are onto something though, the only issue is the fact that 5 will only get hit once a decade if we buy 1000 tickets each time. Maybe keep it 3 digits and pay out every time in the 75-80% range? Roling over the rest. There will still be incentive to cover the board and lock in profit in the amount of the jackpot.

This is actually pretty hard to design. I'll keep thinking.

Keeping it 3 digits will prevent the pot from getting interesting, basically, as you state there at the end. There are other ways of preventing that (like securing 10% of the tickets to myself on opening a draw) but frankly I'd want this to be as independent as possible...

[nelisky]

I just read the New Jersey Pick 3 lottery rules over. They have a simple system:

1) 50% of the ticket sales go to the prize pool
2) (they have a bunch of ways to win, but I'll unitize them all to a number of 'Wins')
    They calculate Ticket intake * .5 / # Wins => Prize Share
3) Each Winning bet gets a prize share.

To compare, the powerball system
A) seeds the jackpot with an amount, ($10mm), advertises twice that as the 'annuitized jackpot', and proceeds as follows:
B) 50% of the ticket sales go to prizes.
C) Roughly 58% of the prizes go to the jackpot, won maybe once a month. The other 42% are for the smaller prizes.


As noted by the powerball website, powerball is designed for people who are motivated by a  big jackpot. Presumably Pick 3 is designed for people who are motivated by a small, frequent (it runs twice a day) payout.

So, either of these seem like fine setups -- hard-headed expected-value-discounted-cashflow-type people only play powerball when the advertised annuity is up in the $250MM+ range, and everyone else plays when they like the sound of the jackpot size. No math people play Pick 3, I'm sure.

I was considering a much simpler 'winner takes it all' setup, as taabl doesn't take absolutely nothing from the lotteries. But that may change, I could make it sponsor a server eventually, in which case a seeded pick 3 with a set prize could be interesting... I'll read over the rules and try to make sense of it

My takeaways:

I) I think your lottery needs to choose a personality: is it going to be a big prize, monthly-ish win type situation, or a frequent small-scale wins one? I don't mean monthly drawings, just likely 'big' wins at once a month.
II) You probably should have a minimum pot for jackpot winners: this ensures people want to play more often. You would fund this out of your own funds to start, and then whatever you decide your rake should be, eventually.
III) An out of the box idea I had is that you could write some sort of rules algorithm and allow anyone to run their own lottery; you'd take, say 10% off the top, and lotto designers could choose their own payouts, jackpot rollovers, etc. Natural selection will tell you the most successful lottery designs quite quickly, I imagine.

I) Not yet, right? We're still playing with possibilities
II) That's what the side pot would be in my latest plan
III) That's actually a great idea! Once the system is stable and running on its own (and on a real server) I'll certainly take a stab at that!

As a sample 'powerball'-esque lottery: if you had a jackpot minimum prize of 10 BTC and two drawings a week, 100 people playing per drawing, odds of winning say 1 in 1000, then you'd see the following:

Week 1, Drawing 1
20 BTC goes in: 10 covers the jackpot: 10 BTC goes to rake
80 BTC goes in: 40 goes to rake, jackpot is now 50.

Week 1, Drawing 2: Jackpot 50
100 BTC goes in: Jackpot is now 100 (no winner)

Week 2, Drawing 1: Jackpot 100
Odds of nobody winning so far: (900 / 1000 ) ^2 = 81%
More people get interested: 150 buy. Nobody wins.

Week 2, Drawing 2: Jackpot 175.
Odds of nobody winning so far: 69%

I don't know if this would be appealing to people or not. I bet if the jackpot were 1,000 BTC, odds of winning say 1 in 5,000 to start then it would be appealing. The simplest way to keep math-oriented speculators out is to keep the expected value down, except in rare  circumstances.

It's gotten too late for me to add more to the conversation, so I'll summarize a TL;DR -- Copy some successful lotteries! Use the rake to provide a guaranteed pool! (Possibly: let people use your software to host their own lotteries).

I copied the local state lottery, but they do two things differently:
- keep 40% of the prize pool for themselves
- Have millions of bets every week

It's very easy to perceive value when you look at a million euro first prize, even though the chances are 13,983,816 to 1, with bet repetitions allowed. But they do prime the lotteries with a big first prize, and only the others are number of bets dependent.

[nelisky]

The lottery is up again!

New rules page, just a basic draft at http://taabl.datlatec.com/rules
5 hex char bets, unclaimed rollover only for 1st prize, all others blend in. 20% of rollover goes to side pot.

Shall we try it this way? I have put 100 in, and anyone wanting to help bootstrap this lottery is thus invited to put an outrageous amount of bets in the system Only 25% will be used for jackpot but still.

[vess]

Okay, I put 20 in. Once confirmed, I'll bet them all.

[vess]

UPDATE: I bet. Then I realized that I could (and should) get total coverage for the last number in the block. Then I learned I can't rescind my bets.

Can I request either a 'full coverage' option, or else just a 'remove this bet' button? I think you had 'remove this bet' at one point.

Also, seeding the lottery with 100 bets of your own is different than putting in a 100 BTC jackpot, by the way. : ). I think it will be more appealing to people if you take a cut, and seed, regardless of the underlying math.

[nelisky]

UPDATE: I bet. Then I realized that I could (and should) get total coverage for the last number in the block. Then I learned I can't rescind my bets.

Can I request either a 'full coverage' option, or else just a 'remove this bet' button? I think you had 'remove this bet' at one point.

I manually gave you your coins back, canceling the tickets. I'll make the coverage based betting soon, and you can then go ahead and bet again.

Also, seeding the lottery with 100 bets of your own is different than putting in a 100 BTC jackpot, by the way. : ). I think it will be more appealing to people if you take a cut, and seed, regardless of the underlying math.

Yep, it's quite different. But I don't want to be an active part on the lottery more than I need to be. It has to run on its own, or it is just too high maintenance. As such I seeded this one with 100 bets, with the promise (as before) that all gains will go right back to the lottery as new bets. But if I quit at any time, because I lost everything or whatever, I need to know all my work wasn't in vain :p

And I still need to find out how I'm going to get this to pay for its own server, but that's for later, when there's enough traffic to justify so.

[FreeMoney]

UPDATE: I bet. Then I realized that I could (and should) get total coverage for the last number in the block. Then I learned I can't rescind my bets.

Can I request either a 'full coverage' option, or else just a 'remove this bet' button? I think you had 'remove this bet' at one point.

I manually gave you your coins back, canceling the tickets. I'll make the coverage based betting soon, and you can then go ahead and bet again.

Also, seeding the lottery with 100 bets of your own is different than putting in a 100 BTC jackpot, by the way. : ). I think it will be more appealing to people if you take a cut, and seed, regardless of the underlying math.

Yep, it's quite different. But I don't want to be an active part on the lottery more than I need to be. It has to run on its own, or it is just too high maintenance. As such I seeded this one with 100 bets, with the promise (as before) that all gains will go right back to the lottery as new bets. But if I quit at any time, because I lost everything or whatever, I need to know all my work wasn't in vain :p

And I still need to find out how I'm going to get this to pay for its own server, but that's for later, when there's enough traffic to justify so.

We/I don't mind if you take the bounty however often you need it.

[nelisky]

We/I don't mind if you take the bounty however often you need it.

Well, thank you But it's all a question of scale. for 2% to cover $20/month (~300BTC) then I'd need to take 75BTC from each weekly draw, which would only happen when we start getting 3750BTC bet pools in!

Of course I can dilute this by putting other services in the same server, but it's a psychological thing, I either see it as paying for itself or as a fun personal project I pay to keep. It's the latter for now, the future may change that.

[FreeMoney]

We/I don't mind if you take the bounty however often you need it.

Well, thank you But it's all a question of scale. for 2% to cover $20/month (~300BTC) then I'd need to take 75BTC from each weekly draw, which would only happen when we start getting 3750BTC bet pools in!

Of course I can dilute this by putting other services in the same server, but it's a psychological thing, I either see it as paying for itself or as a fun personal project I pay to keep. It's the latter for now, the future may change that.

Oh, wow, it'll be awhile before we get that size. Maybe take 5% out of the pool, 2% of a different nice site every time and 3% for you.

Or, take 10% of the excess above like 1000 for a jackpot. So 1200 -> 20; 2000 -> 100; 4000 -> 300

It wouldn't be as regular income, but psychologically it's the least painful place to take it. It's kind of a fallacy, but everyone seems to be happier when big winners do the paying.

[vess]

Thanks for the bet return!

Re: hosting -- I think you could host it on google app engine for free at low traffic volumes.. Of course, you have to like python or java.

[nelisky]

Thanks for the bet return!

Re: hosting -- I think you could host it on google app engine for free at low traffic volumes.. Of course, you have to like python or java.

I love, live and breath python, so that would not be a problem. But I'd still need the bitcoind on another server, and the matter of the fact is that I developed this to learn some new things, and these will take considerable effort to port to app engine. Frankly, I could save some dough, but I'd loose many other advantages, so I don't think I'll take that route.

[nelisky]

Can I request either a 'full coverage' option, or else just a 'remove this bet' button? I think you had 'remove this bet' at one point.

You now have a slightly over complicated version of the 'full coverage' requested. On the bet page you can, in addition to stating how many tickets you are about to buy, define a 'sequence base'. What this does is increment that base sequentially as the bet's rightmost characters. So if you enter '0' as sequence base and '16' as 'number of bets' you'll end up with 16 bets where the rightmost character will go from 0 to F (thus covering 100% of 1 hit) and the other 4 characters random for each bet.

If you do more than 16 for a single char base, it will continue throughout the 2 char sequence (10, 11, 12, etc) and if you enter 5 chars as base, there will be no random part, it will just be a sequence, thus '00000' and '16' will do '00000' through '0000F'.

I hope that helps everyone's betting pleasure!

A small caveat, though. If any of the random bets is already taken the system does not try another for that particular point of the sequence, it will just ignore that one and do the others, thus you can end up with holes and less than requested tickets, but you will get a description of which bets failed so you can do just those. If you want I can fix that, I'm just feeling lazy, really.

[vess]

Awesome! I'm in and sequenced. Thanks!

[nelisky]

Ok, another one is done, this time it was a slow one. I was in for 116 bets, got 10 x 1 hitters. 3 other betters got 1 x 1 hits each.

There's a little side pot now, and a little jackpot too. The jackpot, I just realized, may actually shrink between draws, as it is divided amongst all prizes. The division is good to make betting more interesting, but this may need to change eventually.

Anyway, I am doing final tests on the pick 3 system, which I'll describe here soon, to see if everyone is in agreement with the rules and prizes.

[nelisky]

The BitPick

Still an early version, lots to do, but functionally should be good. Here's how it is set up: I put a value, some value, in to bootstrap the draw. First time around, it's 100BTC. This is the initial pot for whoever guesses the 3 rightmost chars on one of the following blocks.

- No dups, maximum 4096 bets, sure to have a winner in the next block
- Bets are only activated after 6 blocks from the time they are submitted, but that bet is immediately removed from the available pool
- Only one prize, only one winner. Every block from the start of the BitPick is checked (even if the server goes offline for whatever reason, the block order is always maintained)
- Not yet coded in, but any bet not yet active at the time of the winning block gets returned to the better, i.e., that value is not part of the prize pool
- For every 10 bets in, 5 BTCs are added to the prize pool

I'm taking a 50% cut of the bets, but I'm also advancing a value. The largest Draw was 330BTC, over half of which was mine, so I feel that targetting for break even at 200 bets is realistic for the moment. If for some unseen reason we get a lot more bets than that, well, I'll pay for a real server And I'll put more BTCs as initial pool for the next one (not automated yet, when one ends a new one will have to be created manually by me, which may take a few days at worst).

I'll be updating the server now, so share your thoughts as to what should change and, as always, bet your bithearts out! I didn't put a bounty recipient on this one yet, I don't know if I will at all, but first I need to get a feel as to how interesting this is for you and how lucrative it is for me... My lottery is probably the only one in the whole wide world where the house keeps loosing more and more :p

[FreeMoney]

I bought some pick3 tickets and the pot did not increase, it has been a few hours.

[nelisky]

I bought some pick3 tickets and the pot did not increase, it has been a few hours.

That is expected, seeing as I up the pot by 5BTC every 10 bets. It's a psychological thing, I didn't like the 0.5 or 1BTC increases in my tests, and honestly would go as far as 10 each 20, but I thought I'd be stretching the rope there, there's not enough betters yet.

Good luck!