I have written several times before about Peerflix -- Now that I've started applying some tags as well as categories to my items you can now see all the Peerflix stories using that link -- and the issues behind doing a P2P media trading/loaning system. Unlike my own ideas in this area, Peerflix took a selling approach. You sold and bought DVDs, initially for their own internal currency. It was 3 "Peerbux" for new releases, 2 for older ones, and 1 for bargain bin disks.

That system, however, was failing. You would often be stuck for months or more with an unpopular disk. Getting box sets was difficult. So in December they moved to pricing videos in real dollars. I found that interesting because it makes them, in a way, much closer to a specialty eBay. There are still a lot of differences from eBay -- only unboxed disks are traded, they provide insurance for broken disks and most importantly, they set the price on disks.

One can trade DVDs on eBay fairy efficiently but it requires a lot of brain effort because you must put time into figuring good bid and ask prices for items of inconsequential price. Peerflix agreed that this is probably a poor idea, so they decided to set the prices. I don't know how they set their initial prices, but it may have been by looking at eBay data or similar information. Unfortunately, they priced most disks far too high. Indeed, in many cases one could buy the disks on eBay, with the box, for less than the price they set, and certainly sell for more. This forced them to do a major price correction. Their members were distressed to see DVDs they had just bought for $12 now listed as $6. That meant that for those who, like me, treat it more like a rental, planning to resell all disks after they are watched, the rental cost turned out to be much higher than alternatives like Netflix or Blockbuster. Since all disks became cheaper, however, the $6 disk could still be exchanged for another disk that had also been devalued. (It's an interesting question to work out just where the value "went" during this devaluation. Anybody who decides to cash out will have lost value, others don't realize the loss.)

Now the values adjust each week. Many users have identified a risk in purchasing new releases. As expected, new releases tend to depreciate quickly in the first month. At most video rental stores, new releases cost quite a bit more than old ones, but are also most of the traffic. However, the pricing scheme here, where a new release might drop $5, can result in a very expensive rental. (Peerflix adds cost of $1.38 per transaction in the USA.)

Currently, their prices are fluctuating both up and down. They are increasing the price if they see too much demand, dropping it with very low demand. Many low demand disks are dropped to $3. My personal impression is that at the $3 price point, many users may decide they would rather keep a disk than just get $3 for it, even if they are, like me, not disk-keepers. This will take some disks out of circulation.

The cheaper disks seem to circulate much faster than the more expensive ones because of an anomaly. A normal user will have no credit in their account, instead they will have disks. If they send out a cheap disk, the only thing that can happen is to receive another cheap disk. If they send out an expensive disk, they might receive a low-availability expensive disk but it's far more likely they will get 2 or 3 high-availability cheap disks, even if they have used the ranking system to indicate they prefer the expensive disk. Because cheaper disks are by definition ones of greater availability, the only way to stop this will be to let users explicitly say they will wait and not to use their cash on cheap disks.

As you may know, with Netflix you pay a monthly fee to hold 3 disks and get, in theory, unlimited trades. With Peerflix you pay no monthly fee, and you typically will end up holding far more than 3 disks. Because of that you will hold disks for longer, on average, resulting in greater risk of depreciation. If this gets too high, users will not find Peerflix to be cheaper than Netflix, defeating a large part of its value. Unlike Netflix, when you are done with a disk in Peerflix, you may end up sitting on it for quite some time. Under the old system that could be as long as 6 months! With the new system in theory it will get cheap enough to stimulate demand, but that comes with a cost.

While I continue to feel that most users do not want to spend time worrying about how to price disks in a full auction model, if users see the cost of the system as being based on this, they will do so, even irrationally. One thing that might help would be a pledge that disks will normally depreciate monotonically. (This should be the norm anyway, with special exceptions like the movie winning an award.) This requires more advance planning, and temporarily having prices that may be inconsistent with the whims of demand in a small pool. As the pool grows larger, volatility should decrease.

They also may want to consider allow the users who do wish to obsess on price to enter bid and ask prices on disks. If, for example, you note that "Babel" is currently $15, you might put a bid price on it that is lower, and an owner might put an ask price that is lower as well. This could provide a standard stock market style auction, but then only those who took the effort to put in these bids and asks would get much trading done, which defeats the simplicity goal. Instead, the bids and asks could be used to help measure how prices should move.

For example, if a disk is priced at $15, the system could notice that few people want it at the standard price, but there are many bids at $13. It would then lower the price for everybody, not just the $13 bidders, to $13. Those who were selling at "market" price would now sell at $13, while those with a fixed price would find the disk at an unsellable price. I presume that most users would simply put up their disks at market price, though many people would deliberately put in lower bids on new releases, to indicate they wish to see the new release, but are not so eager that they will eat all the depreciation on it.

This is a difficult problem -- how to have market prices for a commodity where most people do not want to spend mental energy on pricing. Netflix has the advantage that there is no mental energy spent on pricing at all -- other than in deciding to pay the monthly fee. Most video stores have very fixed pricing, though people do debate whether you want a $4 new release or a 99 cent classic release due to price.

My proposal above may not reward the bidders enough. They do the mental work of bidding, but everybody gets the new bid price once it is right. If you put the bidders first in line, ahead of the people just taking the disk at market price, you may slow down trading too much for the market price buyers. I believe that many bidders will do their bid calculations anyway, even if it's an irrational use of time. A bid is a way of saying, "I want to get this disk, but only when it becomes cheaper" which is something users already wish to say in the current system. Setting ask prices is less interesting, since prices should mostly depreciate, and asking more than market price should mostly result in never selling a disk. However, there may well be many people who say, "I want to keep this disk, unless some sucker is willing to pay above market for it" which could be expressed with an ask price -- even an ask price written as "Market plus $2" rather than a fixed amount.

In this event, it becomes possible for those who really want a disk right away to be presented with the option to get it right away with some minor overpayment. This may make sense in many cases, since the price of used disks still easily beats the new price in stores, and the overpayment may still cost less than the rental fee at a video store, especially when you factor in the cost of driving there, and the burden of quick return.

I believe the truth about this can only be learned through experimentation. There has not been much done in the way of high-information-flow markets in very low priced commodities that I am aware of.