TOPIC: Applying bids to multi-item auctions

edwin wrote:

Matthew Hayward wrote:

edwin wrote:

Matthew Hayward wrote:

Wade Schwendemann wrote:

Lequinian wrote:

Matthew Hayward wrote: Hmm. What do you do in this scenario:

S1
3 identical items
A wants 2 at $105
B wants 2 at $100
C wants 1 at $1

And this one:

S2
3 identical items
A wants 2 at $105
B wants 1 at $100
C wants 1 at $1

In both scenarios my approach gives A two items, B one item, and both A and B pay $1.

I suspect deviating from my approach gives a large difference in the cost someone pays between S1 and S2.

I suspect this deviation would cause bidders to prefer my approach.

In S1, I believe A has 2 @ $100 and B has 1 @ $100. How does your approach allow B to be happy about not getting his 2nd item in scenario 1?

Probably because he only has to pay $1 for the one he does get!

It's a different way of looking at it.

I'm enjoying this exercise, but it definitely does show different priorities.

In this example, were I B, I might wonder why I didnt get 2 items. If the bid were at my max, I would know that I lost because A bid first without having to ask.

Here is how A and B would see things as the auction went on (Scenario 1 above):

After A's Bid:

Item-1 : A - $0
Item-2 : A - $0

After B's Bid:

Item-1 : A-$0
Item-2 : A-$0
Item-3 : B-$0

Right now, B knows that A's bid is higher than theirs (or it is the same and A bid earlier), and that if they want to get 2 items, they need to increase their bid. A and B both know they are the only bidders (Because the price is still $0).

After C's Bid:

Item-1 : A-$1
Item-2 : A-$1
Item-3 : B-$1


All B has to do to get 2 items is increase their bid. But if they bid their max bid right off the bat, they can go to best rest assured that A has bid more - and that there is no reason to be upset about winning less items than someone else who bid more.

I would never participate in an auction run this way. I would not have confidence in the person running it.

I would post what the freak on not winning an item that I significantly bid higher than the winning bid.

Would you feel the same if all through the bidding you could see that you were not at the top of the list, and therefore knew there were others who had bid more?

I would also ask you to seriously consider this scenario:

Here is the scenario:

3 identical items:

A bids 2 at $105
Edwin bids 2 at $100
C bids 1 at $1

Can you provide any outcome for Edwin which is better than winning 1 token at $1 with any auction system you like?

Can you provide any reason to mistrust the person running this auction, but not mistrust someone else who charges you a few bucks under your max bid, and you just have to take on faith that there was another bidder back there behind you?

It's OK if you don't like this style of auction which, again, I did not invent, and is a common way of running multi-item auctions, and which provably has the same outcome in terms of who wins how many, and is guaranteed to charge each and every winning bidder less than or equal to the winning bid in the other systems provided so far.

It is really incredible to me the amount of pushback "pay less for your tokens" is getting on this thread. I'm in the wrong business, clearly.

I would expect to win one item at $1 and one item at $1+bid increment since that is in alignment with my bid. And a better deal for the bidder. It is all on you to satisfy A not getting two items and is not my issue.

You are saying you would expect to win more items than someone who bid more than you. This is not how auctions work.

You are saying you would rather pay $100 for a token than $1 for a token because if gives you peace of mind.

I guess you won't be bidding on auctions I run with this style.

David Zych wrote:

Matthew Hayward wrote: Here is one thing I would want my auctioneer to guarantee me:

1. The price I pay in no way depends on my max bid - it only depends on what other people's max bids are. (Of course the number of items I win can depend on my max bid).

At Auction House Lambda, your price only depends on your max bid in the case where your max bid is both a winning and a losing bid.

If you bid $x each on 17 PYPs and you win all 17, then either you're paying $x because you had to in order to outbid someone else, or you're paying less than $x because you successfully outbid everyone for less.

If you win fewer than 17 of them, you're definitely paying $x, because the only way you can win fewer than 17 is because somebody else successfully outbid you for some of them (and they won't bother to outbid you on an $x one while there's one still going for less than $x).

2. No one who bid less than me will receive an identical item to me at a lesser price.

At Auction House Lambda, no one who bid after you will receive an identical item to you at a lesser price. Some people who bid earlier than you might get an identical item for one bid increment less.

(or some people who bid after you might pay one bid increment more than you -- but never both at once)

This is important because my time is valuable

Here we agree 100%, and Auction House Lambda has you covered; once you've decided on your max bid(s) it is absolutely to your advantage to submit them as early as possible.

I also think it's important that very few people, if any, would actually rather have 17 PyPs at $105 each than 16 at $80. With the difference in cost they could buy a $250 pack from TD, get their 17th PyP, and have money to burn left over. So you should tell your minions to like, be more smart and not take you so literally .

You don't know in advance that those are the possible outcomes! So the going price is $80 right now and you're winning 16? Maybe you'll get all 17 at $85. Or at $90. Your minion doesn't know what will happen until it puts in an actual bid and sees whether the other people's minions keep bidding or not -- just like a live auction in person (or rather, just like 32 simultaneous live auctions in neighboring rooms).

Our inability to not know the possible outcomes does not change the fact that no rational actor could prefer an outcome of paying $1785 for 17 PyPs versus paying $1280 for 16.

I offer an auction system that prevents this outcome. I'm not sure what's not to like about it - but it sure is generating a lot of opposition!

It's on you to give your minion a set of instructions that prioritizes the outcomes you prefer.

My instructions will be to my minions to consider the auction system which minimizes the cost of achieving my objectives.

I'm not sure why you would insist your minions participate in a system which might result in you paying more than in another system in which you would be guaranteed to win the same items.

But also, those are never going to be the possible outcomes in a real auction.

Really? How did you determine this is impossible?

If PYPs are currently going for $80 and $85, then someone else who hasn't bid yet is going to come along and bid them up to a more reasonable price. And if for some reason they don't, the whole auction will fail anyway and nobody will get any PYPs (the mere thought brings tears to my eyes).

David - you've clearly thought a lot about this, so let me make you a challenge you might enjoy.

I maintain between two auction systems A and B that always award the same bidders the same quantities of tokens, that system A is superior to system B if and only if: the price each bidder pays in A is less than or equal to what they would pay in system B.

Given that definition of "superior" as pertains to auction systems, can you come up with:
a. Any auction system.
b. Any bidding scenario.

Which is superior to the one I have provided, which is:

1. Sort all ( Max amount, quantity ) bids by max amount.
2. Allocate winning items to the bidders in descending order of max amount.
3. Set the price equal to that of the first bid that does not win any items?

?

If you don't like that definition of superior, feel free to ignore it! Can you provide any scenario whatsoever, in any auction system of any kind, where every bidder is gets at least a good outcome as they would be in my system, and at least one gets a better one? Feel free to tell us what your definition of a good outcome would be here!

David Zych wrote:

Matthew Hayward wrote:

David Zych wrote: Forking a new thread from
truedungeon.com/forum?view=topic&defaultmenu=141&catid=584&id=250328&start=48#353987
I agree with Wade, and I'll try to fill in the details of why.

Imagine a very faithful, very patient virtual assistant to whom you have given the instructions "Try to win 17 PYPs at the lowest price you can. Don't pay more than $105 each." This assistant uses that information to place actual bids for individual items on your behalf, one at a time. Every bidder has such an assistant, and the assistants do not share knowledge with each other.

In my opinion, the appropriate auction result is the one that is consistent with this model. Of course there are implementation shortcuts available, but to see how the model works, let's walk through all the actual bids which result deterministically from the application of B's bid.

Initially A bids $0.25 on PYP#1-17:

PYP#1-17 A $0.25
PYP#18-32 $0

1. B bids $0.25 on PYP#18-32.

PYP#1-17 A $0.25
PYP#18-32 B $0.25

...snip...

Here is what I think you're doing, is equivalent to can you tell me if it amounts to the same thing?

1. Take all the bid amounts and quantities, and at the end of the bidding period:

2. Run a virtual auction for each identical item sequentially, one at a time, using eBay rules, removing the winning bids from the bid pool at the start of the subsequent auction.

That's definitely not what I'm doing. I am running all 32 auctions at the same time, so that each virtual assistant can place its next actual bid on whichever item happens to be going for cheapest at that moment. None of the auctions is over until all virtual assistants have stopped bidding.

If so, consider what happens in this scenario:

There are 2 identical items available.

A bids first and wants 1 item at $105 or less
B bids second and wants 1 item at $100 or less
C bids third wants 1 item at $1 or less

Great example! Much easier to follow than 32 items.

Here's the step by step breakdown at Auction House Lambda:
1. A bids $0.25 on item 1. A is happy.
2. B bids $0.25 on item 2. B is happy.
At this point A and B are winning one item each for $0.25
3. C bids $0.50 on item 1 (outbidding A).
4. A bids $0.25 on item 2 (displacing B, since early bids win ties).
5. B bids $0.50 on item 2 (outbidding A).
6. A bids $0.50 on item 1 (displacing C, since early bids win ties).
At this point A and B are winning one item each for $0.50
7. C bids $0.75 on item 1 (outbidding A).
8. A bids $0.50 on item 2 (displacing B, since early bids win ties).
9. B bids $0.75 on item 2 (outbidding A).
10. A bids $0.75 on item 1 (displacing C, since early bids win ties).
At this point A and B are winning one item each for $0.75
11. C bids $1 on item 1 (outbidding A).
12. A bids $0.75 on item 2 (displacing B, since early bids win ties).
13. B bids $1 on item 2 (outbidding A).
14. A bids $1 on item 1 (displacing C, since early bids win ties).
At this point A and B are winning one item each for $1.
15. C gives up. A and B win for $1 each.

Hopefully this helps you see my conceptual model. You'll notice that in this simple case, Auction House Lambda and Auction House Beta end up with the same answer.

Yup - now I understand - I have an updated example that shows the problem with Lambda House below (it's already been mentioned in this thread in regard to some other systems).

Now suppose D wants one item at $110. At Auction House Lambda, D will bid both A and B up to $100, one increment at a time (just like C did), then bid $105 on item 1 (outbidding A). A bids $100 on item 2 (displacing B ), and B gives up.

Result: D wins for $105 and A wins for $100.

Auction House Beta would knock D's price down to $100 as well. That's good news (at least in the short term) if you're D. But I see three comparative downsides:

First, if I'm B, this result doesn't sit quite right. Shouldn't somebody who bids after me need to actually pay more than I was willing to pay in order to win the item I wanted?

No! They only need to bid more than you - not pay more than you.

If you're a bidder you'd appreciate that you could just as easily be A, B, or D in this auction- and look for an auction house which offers the best result no matter who you are (e.g. there's no reason to assume "you" are B - you might also be A, or D):

Lambda House:
A - Wins 1 token for $100
B - Wins 0 tokens.
D - Wins 1 token for $105

Beta House:
A - Wins 1 token for $100
B - Wins 0 tokens.
D - Wins 1 token for $100.

There is no difference if you're A or B, but D does better in Beta House. Since you don't know if you'll be bidder A, or B, or D when you place your bid, and since Beta House guarantees that :

i. You'll always win the quantity that you would have won at Lamdba House.
ii. You'll never pay more, and sometimes pay less than you would at Lambda House.

You would be wise to pick Beta house if your goal is to minimize your spend (again - no matter which house you choose you win the same number of items).

That's not a deal-breaker as long as the rules were stated clearly up front, but IMO it's a consideration.

This is pretty funny to me - as the entire reason we're having this discussion is based on people running auctions without stating the rules clearly up front (e.g. saying they are running the auctions "eBay style" when eBay has no multiunit identical item auctions).

Second, Auction House Beta ends up taking in $5 less. Running an auction house is hard work! If there are two generally reasonable sets of rules to choose from, I certainly won't begrudge the auction house choosing the one that nets them a little bit more in the end. I'm grateful there are auction houses available so I don't have to pay $250 for my PYPs.

You can always send the proprietors of Beta house a $5 tip if this is your concern.

Third, Auction House Beta offers less incentive for early bidding compared to Auction House Lambda. Early bidding helps auctions to succeed, and we all (bidders and auction house managers alike) want TD auctions to succeed.

Why do you say that? They both offer an identical incentive to early bidding, which is that you win the bid if two people bid the same amount and quantity can not satisfy both?


Also - consider this scenario:

3 identical items

A bids on 2 at $105
B bids on 2 at $100
C bids on 1 at $1

Lambda house results in: A wins two at $105, B wins 1 at $100 - right?

Beta house results in A wins two at $1, and B wins 1 at $1.

Beta House's outcomes are radically better for the bidders than Lamdba House's in this case (and never worse in any case).

Wade Schwendemann wrote: It comes.to whether you prioritize the person's request on number of items or their (assumed) desire to pay the lowest price, regardless of how many items they say they want.

I'm not sure 100% what this comment is addressing.

In case it's about your system and mine:

In both your system and my system, the same bidders will always win the same quantities.

Your system will sometimes just charge some of them more.

So we both prioritize requests the same way (same winners, same quantities) - we just come to different prices (sometimes).

Matthew Hayward wrote:

edwin wrote:

Matthew Hayward wrote:

edwin wrote:

Matthew Hayward wrote:

Wade Schwendemann wrote:

Lequinian wrote:

Matthew Hayward wrote: Hmm. What do you do in this scenario:

S1
3 identical items
A wants 2 at $105
B wants 2 at $100
C wants 1 at $1

And this one:

S2
3 identical items
A wants 2 at $105
B wants 1 at $100
C wants 1 at $1

In both scenarios my approach gives A two items, B one item, and both A and B pay $1.

I suspect deviating from my approach gives a large difference in the cost someone pays between S1 and S2.

I suspect this deviation would cause bidders to prefer my approach.

In S1, I believe A has 2 @ $100 and B has 1 @ $100. How does your approach allow B to be happy about not getting his 2nd item in scenario 1?

Probably because he only has to pay $1 for the one he does get!

It's a different way of looking at it.

I'm enjoying this exercise, but it definitely does show different priorities.

In this example, were I B, I might wonder why I didnt get 2 items. If the bid were at my max, I would know that I lost because A bid first without having to ask.

Here is how A and B would see things as the auction went on (Scenario 1 above):

After A's Bid:

Item-1 : A - $0
Item-2 : A - $0

After B's Bid:

Item-1 : A-$0
Item-2 : A-$0
Item-3 : B-$0

Right now, B knows that A's bid is higher than theirs (or it is the same and A bid earlier), and that if they want to get 2 items, they need to increase their bid. A and B both know they are the only bidders (Because the price is still $0).

After C's Bid:

Item-1 : A-$1
Item-2 : A-$1
Item-3 : B-$1


All B has to do to get 2 items is increase their bid. But if they bid their max bid right off the bat, they can go to best rest assured that A has bid more - and that there is no reason to be upset about winning less items than someone else who bid more.

I would never participate in an auction run this way. I would not have confidence in the person running it.

I would post what the freak on not winning an item that I significantly bid higher than the winning bid.

Would you feel the same if all through the bidding you could see that you were not at the top of the list, and therefore knew there were others who had bid more?

I would also ask you to seriously consider this scenario:

Here is the scenario:

3 identical items:

A bids 2 at $105
Edwin bids 2 at $100
C bids 1 at $1

Can you provide any outcome for Edwin which is better than winning 1 token at $1 with any auction system you like?

Can you provide any reason to mistrust the person running this auction, but not mistrust someone else who charges you a few bucks under your max bid, and you just have to take on faith that there was another bidder back there behind you?

It's OK if you don't like this style of auction which, again, I did not invent, and is a common way of running multi-item auctions, and which provably has the same outcome in terms of who wins how many, and is guaranteed to charge each and every winning bidder less than or equal to the winning bid in the other systems provided so far.

It is really incredible to me the amount of pushback "pay less for your tokens" is getting on this thread. I'm in the wrong business, clearly.

I would expect to win one item at $1 and one item at $1+bid increment since that is in alignment with my bid. And a better deal for the bidder. It is all on you to satisfy A not getting two items and is not my issue.

You are saying you would expect to win more items than someone who bid more than you. This is not how auctions work.

You are saying you would rather pay $100 for a token than $1 for a token because if gives you peace of mind.

I guess you won't be bidding on auctions I run with this style.

Incorrect assertion.

edwin wrote:

Matthew Hayward wrote:

edwin wrote:

Matthew Hayward wrote:

edwin wrote:

Matthew Hayward wrote:

Wade Schwendemann wrote:

Lequinian wrote:

Matthew Hayward wrote: Hmm. What do you do in this scenario:

S1
3 identical items
A wants 2 at $105
B wants 2 at $100
C wants 1 at $1

And this one:

S2
3 identical items
A wants 2 at $105
B wants 1 at $100
C wants 1 at $1

In both scenarios my approach gives A two items, B one item, and both A and B pay $1.

I suspect deviating from my approach gives a large difference in the cost someone pays between S1 and S2.

I suspect this deviation would cause bidders to prefer my approach.

In S1, I believe A has 2 @ $100 and B has 1 @ $100. How does your approach allow B to be happy about not getting his 2nd item in scenario 1?

Probably because he only has to pay $1 for the one he does get!

It's a different way of looking at it.

I'm enjoying this exercise, but it definitely does show different priorities.

In this example, were I B, I might wonder why I didnt get 2 items. If the bid were at my max, I would know that I lost because A bid first without having to ask.

Here is how A and B would see things as the auction went on (Scenario 1 above):

After A's Bid:

Item-1 : A - $0
Item-2 : A - $0

After B's Bid:

Item-1 : A-$0
Item-2 : A-$0
Item-3 : B-$0

Right now, B knows that A's bid is higher than theirs (or it is the same and A bid earlier), and that if they want to get 2 items, they need to increase their bid. A and B both know they are the only bidders (Because the price is still $0).

After C's Bid:

Item-1 : A-$1
Item-2 : A-$1
Item-3 : B-$1


All B has to do to get 2 items is increase their bid. But if they bid their max bid right off the bat, they can go to best rest assured that A has bid more - and that there is no reason to be upset about winning less items than someone else who bid more.

I would never participate in an auction run this way. I would not have confidence in the person running it.

I would post what the freak on not winning an item that I significantly bid higher than the winning bid.

Would you feel the same if all through the bidding you could see that you were not at the top of the list, and therefore knew there were others who had bid more?

I would also ask you to seriously consider this scenario:

Here is the scenario:

3 identical items:

A bids 2 at $105
Edwin bids 2 at $100
C bids 1 at $1

Can you provide any outcome for Edwin which is better than winning 1 token at $1 with any auction system you like?

Can you provide any reason to mistrust the person running this auction, but not mistrust someone else who charges you a few bucks under your max bid, and you just have to take on faith that there was another bidder back there behind you?

It's OK if you don't like this style of auction which, again, I did not invent, and is a common way of running multi-item auctions, and which provably has the same outcome in terms of who wins how many, and is guaranteed to charge each and every winning bidder less than or equal to the winning bid in the other systems provided so far.

It is really incredible to me the amount of pushback "pay less for your tokens" is getting on this thread. I'm in the wrong business, clearly.

I would expect to win one item at $1 and one item at $1+bid increment since that is in alignment with my bid. And a better deal for the bidder. It is all on you to satisfy A not getting two items and is not my issue.

You are saying you would expect to win more items than someone who bid more than you. This is not how auctions work.

You are saying you would rather pay $100 for a token than $1 for a token because if gives you peace of mind.

I guess you won't be bidding on auctions I run with this style.

Incorrect assertion.

Lol - which one?

Matthew Hayward wrote:

edwin wrote:

Matthew Hayward wrote:

edwin wrote:

Matthew Hayward wrote:

edwin wrote:

Matthew Hayward wrote:

Wade Schwendemann wrote:

Lequinian wrote:

Matthew Hayward wrote: Hmm. What do you do in this scenario:

S1
3 identical items
A wants 2 at $105
B wants 2 at $100
C wants 1 at $1

And this one:

S2
3 identical items
A wants 2 at $105
B wants 1 at $100
C wants 1 at $1

In both scenarios my approach gives A two items, B one item, and both A and B pay $1.

I suspect deviating from my approach gives a large difference in the cost someone pays between S1 and S2.

I suspect this deviation would cause bidders to prefer my approach.

In S1, I believe A has 2 @ $100 and B has 1 @ $100. How does your approach allow B to be happy about not getting his 2nd item in scenario 1?

Probably because he only has to pay $1 for the one he does get!

It's a different way of looking at it.

I'm enjoying this exercise, but it definitely does show different priorities.

In this example, were I B, I might wonder why I didnt get 2 items. If the bid were at my max, I would know that I lost because A bid first without having to ask.

Here is how A and B would see things as the auction went on (Scenario 1 above):

After A's Bid:

Item-1 : A - $0
Item-2 : A - $0

After B's Bid:

Item-1 : A-$0
Item-2 : A-$0
Item-3 : B-$0

Right now, B knows that A's bid is higher than theirs (or it is the same and A bid earlier), and that if they want to get 2 items, they need to increase their bid. A and B both know they are the only bidders (Because the price is still $0).

After C's Bid:

Item-1 : A-$1
Item-2 : A-$1
Item-3 : B-$1


All B has to do to get 2 items is increase their bid. But if they bid their max bid right off the bat, they can go to best rest assured that A has bid more - and that there is no reason to be upset about winning less items than someone else who bid more.

I would never participate in an auction run this way. I would not have confidence in the person running it.

I would post what the freak on not winning an item that I significantly bid higher than the winning bid.

Would you feel the same if all through the bidding you could see that you were not at the top of the list, and therefore knew there were others who had bid more?

I would also ask you to seriously consider this scenario:

Here is the scenario:

3 identical items:

A bids 2 at $105
Edwin bids 2 at $100
C bids 1 at $1

Can you provide any outcome for Edwin which is better than winning 1 token at $1 with any auction system you like?

Can you provide any reason to mistrust the person running this auction, but not mistrust someone else who charges you a few bucks under your max bid, and you just have to take on faith that there was another bidder back there behind you?

It's OK if you don't like this style of auction which, again, I did not invent, and is a common way of running multi-item auctions, and which provably has the same outcome in terms of who wins how many, and is guaranteed to charge each and every winning bidder less than or equal to the winning bid in the other systems provided so far.

It is really incredible to me the amount of pushback "pay less for your tokens" is getting on this thread. I'm in the wrong business, clearly.

I would expect to win one item at $1 and one item at $1+bid increment since that is in alignment with my bid. And a better deal for the bidder. It is all on you to satisfy A not getting two items and is not my issue.

You are saying you would expect to win more items than someone who bid more than you. This is not how auctions work.

You are saying you would rather pay $100 for a token than $1 for a token because if gives you peace of mind.

I guess you won't be bidding on auctions I run with this style.

Incorrect assertion.

Lol - which one?

As stated.

Just confirm that you auction will preclude people in California from participating like your ebay auctions?

edwin wrote:

Matthew Hayward wrote:

edwin wrote:

Matthew Hayward wrote: You are saying you would expect to win more items than someone who bid more than you. This is not how auctions work.

You are saying you would rather pay $100 for a token than $1 for a token because if gives you peace of mind.

I guess you won't be bidding on auctions I run with this style.

Incorrect assertion.

Lol - which one?

As stated.

Just confirm that you auction will preclude people in California from participating like your ebay auctions?

Let's back up, because I have no idea where you're going with this.

Let me try again.

You get to resolve this auction however you want. The scenario is:

3 identical items.

Bidder A bids $105 on 2 of them.
Bidder B bids $100 on 2 of them.
Bidder C bids $1 on 1 of them.

Describe any allocation you like of winning items and bidding amounts that you think is fair and reasonable, and which you would agree is fair and reasonable if you were Bidder A, or bidder B, or bidder C.

Then we'll compare it with my approach.

My guarantee to you is that my approach will lead to the same or better outcome for all of Bidder A, B, and C, as measured by how much they pay for their won items, and that they will all win the same quantities as they would in your scenario.

Matthew Hayward wrote:

Wade Schwendemann wrote: It comes.to whether you prioritize the person's request on number of items or their (assumed) desire to pay the lowest price, regardless of how many items they say they want.

I'm not sure 100% what this comment is addressing.

In case it's about your system and mine:

In both your system and my system, the same bidders will always win the same quantities.

Your system will sometimes just charge some of them more.

So we both prioritize requests the same way (same winners, same quantities) - we just come to different prices (sometimes).

I think what I meant to say is, if B wanted 1 PyP at 105 and a 2nd only at a lower price point, or two of and only if their sum total is <X, that should be the bid. Whether the auctioneer will accept such a bid is a different story.

I agree that all of your points result in at least as good of a deal and sometimes a better deal for some bidders than the way I have run my auctions.

Speaking strictly for myself, I believe that the way that I have done them in the past is fair to the bidders, and I will very likely continue to run them that way.

Every bid I receive generates more work for me to compile and update. I have absolutely no technical skill, so that has to be factored into my running of an auction. I do genuinely want to have as much incentive as possible for people to bid early, with their true max. If paying one bid increment less for an item helps that happen, I am all for it.

I also hope I don't lose any bidders as a result of this.

Wade Schwendemann wrote:

Matthew Hayward wrote:

Wade Schwendemann wrote: It comes.to whether you prioritize the person's request on number of items or their (assumed) desire to pay the lowest price, regardless of how many items they say they want.

I'm not sure 100% what this comment is addressing.

In case it's about your system and mine:

In both your system and my system, the same bidders will always win the same quantities.

Your system will sometimes just charge some of them more.

So we both prioritize requests the same way (same winners, same quantities) - we just come to different prices (sometimes).

I think what I meant to say is, if B wanted 1 PyP at 105 and a 2nd only at a lower price point, or two of and only if their sum total is <X, that should be the bid. Whether the auctioneer will accept such a bid is a different story.

I agree that all of your points result in at least as good of a deal and sometimes a better deal for some bidders than the way I have run my auctions.

Speaking strictly for myself, I believe that the way that I have done them in the past is fair to the bidders, and I will very likely continue to run them that way.

Every bid I receive generates more work for me to compile and update. I have absolutely no technical skill, so that has to be factored into my running of an auction. I do genuinely want to have as much incentive as possible for people to bid early, with their true max. If paying one bid increment less for an item helps that happen, I am all for it.

I also hope I don't lose any bidders as a result of this.

If the discussion so far is anything to go on, they shall flock to you and away from me .

My interest in this got going when I noticed:

Hey... wait-a-minute - sometimes I could get a better result in these auctions by bidding less!

You really don't want to be the top bidder in these auctions as they are run - you want to be the lowest bidder who wins the desired number of tokens. E.g. if you are going to win X PyPs at the end of the day when all bids are considered and your max bid is $Y - by increasing your max bid you could potentially end up paying more (possibly a lot more).

Which goes against what we agree on: an auction where the incentives align on putting in your top dollar early.

I have run auctions the way you're doing it (or similar) in the past - I think it's fine - I just think I now have a way that is more friendly to bidders that I'm excited to try out.

If the feedback here is any guide I'll be branded a shady operator whose ripping people off when someone doesn't win their full request at a price less than their max bid - so I've got that to look forward to .

Matthew Hayward wrote:

David Zych wrote: If I'm B, this result doesn't sit quite right. Shouldn't somebody who bids after me need to actually pay more than I was willing to pay in order to win the item I wanted?

No! They only need to bid more than you - not pay more than you.

I think this is the key point on which we disagree.

3 identical items

A bids on 2 at $105
B bids on 2 at $100
C bids on 1 at $1

Lambda house results in: A wins two at $105, B wins 1 at $100 - right?

No, A wins two at $100 and B wins one at $100; the bid never gets to $105 because A has no reason to increase it that high and B isn't willing to. (This doesn't alter the point you're trying to make, but it's disconcerting to me that you still don't seem to understand the system that you're arguing so strongly against.)

Beta house results in A wins two at $1, and B wins 1 at $1.
Beta House's outcomes are radically better for the bidders than Lamdba houses.

Except it leaves B dissatisfied, because he wanted a second item and was perfectly willing to pay far more for it than the price for which it actually sold. B must take it entirely on faith that he lost out on the second item because the secret max bid of A which was never realized was higher than the secret max bid of B which was also never realized. Moreover, B must take it entirely on faith that A (whose identity he doesn't even know) truly was willing to pay that much.

Taking this one step further, B has the option of rebidding $150 in order to win two items at $1 rather than one item at $1 -- even if he's not actually willing to pay $150. Of course he might get caught in his lie, if multiple other bidders come along and bid exactly $150, but maybe the chance of getting away with it is a chance he's willing to take at Beta House.

In Lambda House, every bidder who lost an item can see exactly how and why they lost: someone else is actually paying at least as much as they were willing to pay, so if their max bid was honest then they have no regrets about losing. And if for some reason A fails to actually cough up the promised $100, the auctioneer can then offer the item to B instead. But that doesn't happen very often at Lambda House, because there is never any incentive to lie about your max bid.

Auction House Beta offers less incentive for early bidding compared to Auction House Lambda. Early bidding helps auctions to succeed, and we all (bidders and auction house managers alike) want TD auctions to succeed.

Why do you say that? They both offer an identical incentive to early bidding, which is that you win the bid if two people bid the same amount and quantity can not satisfy both?

No, they do not offer identical incentive. As my previous examples have shown, earlier bidders at Lambda House sometimes end up paying a final price that is one bid increment less than a later bidder. That can never happen at Beta House, hence less incentive.

Can you provide any scenario whatsoever, in any auction system of any kind, where every bidder is gets at least a good outcome as they would be in my system, and at least one gets a better one?

Absolutely. At Quarter House, the winning price for every item is always $0.25 regardless of the max bid amounts. In this post's example, A wins two items at $0.25, and B wins one item at $0.25. Everyone saves money! By your stated definition of superior, Quarter House is the best auction system.

Obviously I don't think Quarter House is a good idea; it has a number of fundamental flaws. In my opinion, Beta House has several of the same flaws as Quarter House, albeit to a lesser degree. My point here is that you have to allow for some other considerations besides just minimal prices, or you end up championing Quarter House.

Fun fact: the current winning bid for a PYP at Quarter House is "infinity squared plus seven". But nobody knows that, so they keep trying and failing to outbid it with "infinity times three".

So David already said what I was basically going to say about transparency in the system.
However, I feel it important to point out that:

Auction house Lambda treats these two encounters the exact same way:

A bids $105 on 2
B bids $100 on 1
B bids $100 on 1
C bids $1 on 3

and

A bids $105 on 2
B bids $100 on 2
C bids $1 on 3


However the system Matthew is describing see B's as separate bids in one instance, (one that wins something and one that doesn't), but one single bid that did win something in another, and thus results in vastly different outcomes.

I see that as an issue. I would expect my bids to show me in a quite straightforward manner where I stand as soon as an update occurs. I also worry with the system Matthew is describing that it give me a RIDICULOUS amount of "f - you" power as a bidder.

So I am B, and I put in my bids as we have been saying, and now I see this:

item 1: A @ $1
item 2: A @ $1
item 3: B: @ 1$

Cool, I understand the system, and know that A must have beat my bids of $100.

You know what? Screw that guy. I send Matthew a bid of $100. I know for a fact A bid more than that. Now that is the first bid that isn't winning anything (I know it will up all of our bids, but it will up more of his!)
Now all bids are at $100.

And that also doesn't account for the fact that I am sending in a bid of $100, that I have already sent in, and now it is going to have a different effect. That seems strange. I understand how Matthew's proposed system works, but it seems odd to me that me sending in the exact same bid 5 minutes apart should have such a drastically different outcome.