6 ms·

I actually used to write book arbitrage software, and this person has managed to completely misunderstand it. I'm not arguing here for its societal value, just

I actually used to write book arbitrage software, and this person has managed to completely misunderstand it. I'm not arguing here for its societal value, just that dan appears to have misunderstood some fundamental bits about how it works. > Robot 1 offers Book A for $10. It probably doesn't have it in stock, but it knows where it can buy it for, say, $9, drop-ship it to the customer without ever seeing the acutal product, and make a buck on the arbitrage. Robot 2 notices this and lists the same book for $11, hoping to do the same thing. Robot 1 notices Robot 2's listing, doesn't know or care that Robot 2 is another arbitrage trader, and reprices its listing for the book to $12. Robot 2 notices this and goes to $13, and away we go. This makes no sense; the fundamental rule of amazon bots is that consumers buying used books almost always buy the lowest-priced used book available. So the bot war would send the price down, not up. Why in the world would bot 2 offer the book for $11? It would know that nobody would buy it when there was one already listed for $10. The silly-high-priced books that I saw when I was working for the book company were rare long-tail books. One of the discoveries that many people have made is that for rare books, whether or not you understand what the heck it is, some % of them will get bought even if they're listed for silly high prices. Enough people will pay thousands of dollars (not an exaggeration) for a used copy of a book that they can't find elsewhere that stocking the long tail of books can in fact be very valuable. That means that, when you find lots of rare books, your incentive is to throw out some crazy-high price for each, because some % of them will, in fact, get paid, making the whole enterprise worthwhile. edit: Arbitrage on the same market Does Not Work, in my experience, and to just claim that it does with no evidence doesn't make it any more likely to do so.

JetSpiegel 13y ago
Bot 2 sells higher because it needs to buy the book from Bot 1. It's either in the article on in the links inside.
- nickff 13y ago
  Why does it need to buy the book at all?
  - mcguire 13y ago
    Because someone paid the owner of Bot2 for it.
- llimllib 13y ago
  that doesn't work on amazon, any book with an ISBN will get condensed into one listing. You can get a little tricky with it, but not enough to allow intra-market arbitrage. I understand that he's claiming it exists, and it might with other products, but my experience was that it does not with books.
  - JetSpiegel 13y ago
    How doesn't it work? The buyer buys more expensive because the seller might have a better feedback score.
    - llimllib 13y ago
      Buyers are willing to pay a very small premium for feedback score. Furthermore, one of the bots will know that it has a lower feedback score, immediately killing any hypothetical positive feedback loop.
jbenz 13y ago
>Why in the world would bot 2 offer the book for $11? It would know that nobody would buy it when there was one already listed for $10. The author covers that point with this quote: >In this case, buyers could easily find the "original" products for themselves, and pay less... but the search results are cluttered with millions-sold arbitrage super-stores
- llimllib 13y ago
  that doesn't seem true to me, and it certainly wasn't when I was writing the software. I follow what you're saying, but it's not what happened IME.
mcguire 13y ago
Our Oriental Heritage, by Will Durant currently has about 56 used copies for sale on Amazon. It's not especially rare; I bought it to fill out a half-set I bought from an antique store for $20. But the two highest prices are $225.00 and $588.99 when there are seven listings under $3.00. That's just one example that I noticed the other day. Presumably, the cheaper copies are either poor condition or from high-volume stores. The middle prices seem to be in better condition, from lower-volume sellers. The higher prices are from sellers with many different levels of volume. Presumably, they're hoping for the inattentive-buyer market.
- llimllib 13y ago
  My guess would be that they listed the book a long time ago and lost track of it or that their pricing bot has a bug. Or they're humans (one dedicated person can do a surprising amount of book sales, we discovered). Our experience was that there is no inattentive-buyer market. Those high-priced books will not sell. edit: I'd add that we didn't like to stock books with low prices like that. There's no margin in the <$5+s&h books you see on Amazon. Instead, we had a big pulper and we'd pulp our 10m copies of "Patriot Games" and sell them to recyclers. My understanding is that my former employers have moved to start making recycled paper themselves and selling that these days.
  - morsch 13y ago
    Seems like an appropriate name for that machine. snicker
sneak 13y ago
> That means that, when you find lots of rare books, your incentive is to throw out some crazy-high price for each, because some % of them will, in fact, get paid, making the whole enterprise worthwhile. Oh, so it's just like domain squatting. Got it.
- llimllib 13y ago
  What is the correct value to sell a book for, when you know it's rare, and know nothing else about it?
- tbrownaw 13y ago
  How so? For the rare books, I expect the price would be limited both by value and by how much time it would take to find another copy somewhere (say, a university library two towns away) and how much the buyer's time is worth. Domain squatting I thought tends to be more about trademarks and trying to confuse people.
- bmm6o 13y ago
  That seems like an excessively negative analogy, and not especially apt given that these are books that were presumably published in some quantity.
ableal 13y ago
> Why in the world would bot 2 offer the book for $11? Dan's explanation: "The price-setting robot sets its price far enough above that of the other vendor that if someone buys, the robot-owner can buy the book from the other vendor, send it to the customer, and still make money." To make money - out of nothing except a listing - only requires a buyer, either misguided, malicious or unfortunate. This might even work in large numbers by pure chance. Or it might not, but someone is trying. > consumers buying used books almost always buy the lowest-priced There it is, a bet on the complement of "almost always", at an insignificant cost.
- llimllib 13y ago
  > The price-setting robot sets its price far enough above that of the other vendor that if someone buys, the robot-owner can buy the book from the other vendor, send it to the customer, and still make money. Yes, this is what book arbitrage is, I hope I'm aware of the concept! It's what I did for a few years. > To make money - out of nothing except a listing - only requires a buyer, either misguided, malicious or unfortunate. Or none of the above! If they buy it on Amazon for $10, but it was only $5 on AbeBooks, and they never would have found it there, they're a perfectly happy customer. > There it is, a bet on the complement of "almost always", at an insignificant cost. I'm suggesting that, as a person with several years professional experience doing so, nobody makes money on Amazon by being anything except the lowest-priced used book available. I'm further suggesting that the bot wars that happen (often!) are negative feedback loops, causing prices to drop, not the other way around, and he has provided neither examples nor a reasonable explanation of why bots would fight to be lower on the list. Being first on the list of used books is everything when selling on Amazon, and if there are bots going the opposite way, they won't last long. Low cost ≠ free.
tzs 13y ago
> This makes no sense; the fundamental rule of amazon bots is that consumers buying used books almost always buy the lowest-priced used book available. So the bot war would send the price down, not up Consumers often take into account seller reputation. If a book is $20 at a seller with stellar feedback from a large number of customers, and $18 at a seller with very little feedback, I'll probably buy from the $20 seller. What happened in the case with the two bots on Amazon was that one bot was pricing for a seller with a stellar reputation based on a lot of feedback. It was finding the low price from sellers whose reputations were not as good, and setting the price enough above that to make a profit reselling from the low seller if someone ordered. The low seller's bot was using a strategy of undercutting the lowest other seller. The amount the first bot was adding was more than the amount the second bot was undercutting by, so the net result was the price zoomed up.
auctiontheory 13y ago
This has been studied and documented in detail, about a year ago - five minutes on google will find it. Basically, someone tracked the two prices for a single title rising in lock step - just the thing you say makes no sense. My own book, which is available new from Amazon for around $12, has listed for several hundred thousand dollars from a reseller.
- chc 13y ago
  Just because something happens sometimes doesn't mean it's par for the course. That guy selling your book for $100k? He's probably not selling your book at all. It's hard to believe nobody would have worked out that this is not the most efficient scheme.
mcpherrinm 13y ago
You wrote the bots, and I wrote the code that (tries) to keep them from messing with Amazon (as an intern). People on Amazon don't just buy the lowest priced item: They buy from whoever won the "buy box": The big "add to cart"/"buy now 1 clock" buttons. You seem to be talking about selling used books, where the market is a little different than "new" items. Price is obviously the biggest determination of who wins, but you can be disqualified because you don't have a high enough ranking, and other factors related to customer satisfaction. So an established merchant may win the buy box even if they're 50 cents higher. And that's exactly where this scheme comes into play, as soon as you've got two of these bots playing off each other and no other sellers qualified to win the buy box.
- llimllib 13y ago
  Neat! We were also obsessed with our customer rating, that just didn't happen to be my job. It was tough to get happy customers to come back and give us 5-star ratings; people with bad experiences would come by and rate but those with perfect experiences never would. Or people who got a book in perfect condition, on time, would rate us 3-stars. What the heck did they expect? Obviously, shipping on-time and good product is crucial too. Again, mostly just not my job. So we were definitely working on the customer satisfaction angle as well, there just wasn't much that I could do about it, so I talk less about it.

JetSpiegel 13y ago

Bot 2 sells higher because it needs to buy the book from Bot 1. It's either in the article on in the links inside.

nickff 13y ago
Why does it need to buy the book at all?
- mcguire 13y ago
  Because someone paid the owner of Bot2 for it.
llimllib 13y ago
that doesn't work on amazon, any book with an ISBN will get condensed into one listing. You can get a little tricky with it, but not enough to allow intra-market arbitrage. I understand that he's claiming it exists, and it might with other products, but my experience was that it does not with books.
- JetSpiegel 13y ago
  How doesn't it work? The buyer buys more expensive because the seller might have a better feedback score.
  - llimllib 13y ago
    Buyers are willing to pay a very small premium for feedback score. Furthermore, one of the bots will know that it has a lower feedback score, immediately killing any hypothetical positive feedback loop.

jbenz 13y ago

>Why in the world would bot 2 offer the book for $11? It would know that nobody would buy it when there was one already listed for $10. The author covers that point with this quote: >In this case, buyers could easily find the "original" products for themselves, and pay less... but the search results are cluttered with millions-sold arbitrage super-stores

llimllib 13y ago
that doesn't seem true to me, and it certainly wasn't when I was writing the software. I follow what you're saying, but it's not what happened IME.

mcguire 13y ago

Our Oriental Heritage, by Will Durant currently has about 56 used copies for sale on Amazon. It's not especially rare; I bought it to fill out a half-set I bought from an antique store for $20. But the two highest prices are $225.00 and $588.99 when there are seven listings under $3.00. That's just one example that I noticed the other day. Presumably, the cheaper copies are either poor condition or from high-volume stores. The middle prices seem to be in better condition, from lower-volume sellers. The higher prices are from sellers with many different levels of volume. Presumably, they're hoping for the inattentive-buyer market.

llimllib 13y ago
My guess would be that they listed the book a long time ago and lost track of it or that their pricing bot has a bug. Or they're humans (one dedicated person can do a surprising amount of book sales, we discovered). Our experience was that there is no inattentive-buyer market. Those high-priced books will not sell. edit: I'd add that we didn't like to stock books with low prices like that. There's no margin in the <$5+s&h books you see on Amazon. Instead, we had a big pulper and we'd pulp our 10m copies of "Patriot Games" and sell them to recyclers. My understanding is that my former employers have moved to start making recycled paper themselves and selling that these days.
- morsch 13y ago
  Seems like an appropriate name for that machine. snicker

sneak 13y ago

> That means that, when you find lots of rare books, your incentive is to throw out some crazy-high price for each, because some % of them will, in fact, get paid, making the whole enterprise worthwhile. Oh, so it's just like domain squatting. Got it.

llimllib 13y ago
What is the correct value to sell a book for, when you know it's rare, and know nothing else about it?
tbrownaw 13y ago
How so? For the rare books, I expect the price would be limited both by value and by how much time it would take to find another copy somewhere (say, a university library two towns away) and how much the buyer's time is worth. Domain squatting I thought tends to be more about trademarks and trying to confuse people.
bmm6o 13y ago
That seems like an excessively negative analogy, and not especially apt given that these are books that were presumably published in some quantity.

ableal 13y ago

> Why in the world would bot 2 offer the book for $11? Dan's explanation: "The price-setting robot sets its price far enough above that of the other vendor that if someone buys, the robot-owner can buy the book from the other vendor, send it to the customer, and still make money." To make money - out of nothing except a listing - only requires a buyer, either misguided, malicious or unfortunate. This might even work in large numbers by pure chance. Or it might not, but someone is trying. > consumers buying used books almost always buy the lowest-priced There it is, a bet on the complement of "almost always", at an insignificant cost.

llimllib 13y ago
> The price-setting robot sets its price far enough above that of the other vendor that if someone buys, the robot-owner can buy the book from the other vendor, send it to the customer, and still make money. Yes, this is what book arbitrage is, I hope I'm aware of the concept! It's what I did for a few years. > To make money - out of nothing except a listing - only requires a buyer, either misguided, malicious or unfortunate. Or none of the above! If they buy it on Amazon for $10, but it was only $5 on AbeBooks, and they never would have found it there, they're a perfectly happy customer. > There it is, a bet on the complement of "almost always", at an insignificant cost. I'm suggesting that, as a person with several years professional experience doing so, nobody makes money on Amazon by being anything except the lowest-priced used book available. I'm further suggesting that the bot wars that happen (often!) are negative feedback loops, causing prices to drop, not the other way around, and he has provided neither examples nor a reasonable explanation of why bots would fight to be lower on the list. Being first on the list of used books is everything when selling on Amazon, and if there are bots going the opposite way, they won't last long. Low cost ≠ free.

tzs 13y ago

> This makes no sense; the fundamental rule of amazon bots is that consumers buying used books almost always buy the lowest-priced used book available. So the bot war would send the price down, not up Consumers often take into account seller reputation. If a book is $20 at a seller with stellar feedback from a large number of customers, and $18 at a seller with very little feedback, I'll probably buy from the $20 seller. What happened in the case with the two bots on Amazon was that one bot was pricing for a seller with a stellar reputation based on a lot of feedback. It was finding the low price from sellers whose reputations were not as good, and setting the price enough above that to make a profit reselling from the low seller if someone ordered. The low seller's bot was using a strategy of undercutting the lowest other seller. The amount the first bot was adding was more than the amount the second bot was undercutting by, so the net result was the price zoomed up.

auctiontheory 13y ago

This has been studied and documented in detail, about a year ago - five minutes on google will find it. Basically, someone tracked the two prices for a single title rising in lock step - just the thing you say makes no sense. My own book, which is available new from Amazon for around $12, has listed for several hundred thousand dollars from a reseller.

chc 13y ago
Just because something happens sometimes doesn't mean it's par for the course. That guy selling your book for $100k? He's probably not selling your book at all. It's hard to believe nobody would have worked out that this is not the most efficient scheme.

mcpherrinm 13y ago

You wrote the bots, and I wrote the code that (tries) to keep them from messing with Amazon (as an intern). People on Amazon don't just buy the lowest priced item: They buy from whoever won the "buy box": The big "add to cart"/"buy now 1 clock" buttons. You seem to be talking about selling used books, where the market is a little different than "new" items. Price is obviously the biggest determination of who wins, but you can be disqualified because you don't have a high enough ranking, and other factors related to customer satisfaction. So an established merchant may win the buy box even if they're 50 cents higher. And that's exactly where this scheme comes into play, as soon as you've got two of these bots playing off each other and no other sellers qualified to win the buy box.

llimllib 13y ago
Neat! We were also obsessed with our customer rating, that just didn't happen to be my job. It was tough to get happy customers to come back and give us 5-star ratings; people with bad experiences would come by and rate but those with perfect experiences never would. Or people who got a book in perfect condition, on time, would rate us 3-stars. What the heck did they expect? Obviously, shipping on-time and good product is crucial too. Again, mostly just not my job. So we were definitely working on the customer satisfaction angle as well, there just wasn't much that I could do about it, so I talk less about it.