Selling information with bitcoins

[grondilu]

Here are some of my thoughts concerning the problem of selling information at distance.

As you know, selling information is a hot topic nowadays.  Since the popular rise of worldwide P2P filesharing networks, major audiovisual entertainment companies have suffered from a dramatic loss of profit.  Thanks to intense lobying towards politics, they're slowly managing to have some laws voted in order to protect the so-called "intellectual property", mainly via internet control and regulation of ISPs.

Now, I won't discuss here about intellectual property and why it should or not be protected using public force.  Instead, I'd like to suggest a way to sell information from a purely technical basis, without any need of juridical intervention or coercion.

Selling information is tricky.  This is old news.  Several professions have to deal with this problem on a daily basis.  Journalists, for instance, must be very carefull not to divulgate their *scoops* before it is printed and available for sell on main street.  This illustrates the very point about information :  it is yours as long as it is a secret.  As soon as you communicate it to just one person, it is not yours anymore.  To sell information, you must sell divulgation, not diffusion.

Therefore, the main idea about selling information is to sell it once, not to try to sell several copies of it.  Of course, this means that you have to sell it to a very high price, at least as high as your production cost, more if you want to make profit.   Selling such an expensive product is almost impossible if you want to sell it to the general public.  A solution is to sell your information by parts.  One byte at a time.

Such a decomposition of selling would require an easy an efficient micropaiement solution, for which I think Bitcoin is an excellent candidate.

There are several ways to imagine how such a selling could occur.  I'd like to introduce one, which would be based on a hash tree decomposition of the information.

Say I want to sell a 1Go file called foobar.avi.

First I create a sha-256 tree of this file, with 32Ko sized blocks.  Then, I make sure I have a way to designate each of thes hashes in the tree in a unique way.  The root hash is just a other name for the file "foobar.avi".  It is a hash of a 32Ko file containing 32Ko/256= 128 other hashes.  Each of them also are hashes of 128 other hashes.  Thus, I can easily identify each hash using a big integer in a 128-base notation, for instance :

109.77.56

Since a 1Go file contains at most 32768 32ko files, then the number will have at most 3 digits (in a 128-base notation, though).

So if we add the root hash as a prefix, we have a precise way to designate any hash of the hash tree :

faebc3c13a28521ab38d2482bae4e27e7c754f665f03cebdf7917f34864676bf:109.77.56

where faeb... is the root hash.

Now customers can contact me and ask for a 32Ko block whose hash-256 is referenced as above.  I'll sell it to them, at a price that will be about 32768 less than the total earnings I expect to get from foobar.avi.  Customers can not view foobar.avi with only one block, but at least they can be asured that they bought a part of the hash tree (provided that they already own "upper" blocks).

Customers then share their blocks on a P2P network, or any other way, it's not my problem.  But they
will only be able to watch foobar.avi once allblocks will be sold, and my money raised.

[mizerydearia]

This is quite a brilliant idea and could be used by mafiaa if they had least bit of intelligence

[grondilu]

This is quite a brilliant idea and could be used by mafia if they had least bit of intelligence

Well, my guess is porn industry will be one of the first to do something like that.

Apart from this, I realize that the base-128 designation system is not necessary.

Customers first buy the root hash block, then each of hashes it is made of, and so on, until what they get is identified to be in the expected format (avi for instance).

Also, seller should estimate the number of potential buyers in order to chose an appropriate block size (and thus the correct number of blocks).   This is necessary to adapt the price of each block.

[theymos]

This could be used in a modified BitTorrent client where the initial seed charges 0.1 BTC per piece or whatever. Every peer just has to load up their client with 50 BTC and start downloading from the seed and those peers that will share. Maybe the peers could charge a small amount, too.

[grondilu]

This could be used in a modified BitTorrent client where the initial seed charges 0.1 BTC per piece or whatever. Every peer just has to load up their client with 50 BTC and start downloading from the seed and those peers that will share. Maybe the peers could charge a small amount, too.

I'm not sure it would be that easy.  I doubt current P2P protocols distribute files in a tree hash form.  A tree hash form is necessary in order to make sure you don't buy rubbish.

Maybe the GNUnet protocol would be compatible.

[mizerydearia]

This could be used in a modified BitTorrent client where the initial seed charges 0.1 BTC per piece or whatever. Every peer just has to load up their client with 50 BTC and start downloading from the seed and those peers that will share. Maybe the peers could charge a small amount, too.

I'm not sure it would be that easy.  I doubt current P2P protocols distribute files in a tree hash form.  A tree hash form is necessary in order to make sure you don't buy rubbish.

Maybe the GNUnet protocol would be compatible.

http://en.wikipedia.org/wiki/Hash_tree#Tiger_tree_hash ?

Tiger tree hashes are used in the Gnutella, Gnutella2, and Direct Connect P2P file sharing protocols and in file sharing applications such as Phex, BearShare, LimeWire, Shareaza, DC++[4] and Valknut

[grondilu]

This could be used in a modified BitTorrent client where the initial seed charges 0.1 BTC per piece or whatever. Every peer just has to load up their client with 50 BTC and start downloading from the seed and those peers that will share. Maybe the peers could charge a small amount, too.

I'm not sure it would be that easy.  I doubt current P2P protocols distribute files in a tree hash form.  A tree hash form is necessary in order to make sure you don't buy rubbish.

Maybe the GNUnet protocol would be compatible.

http://en.wikipedia.org/wiki/Hash_tree#Tiger_tree_hash ?

Tiger tree hashes are used in the Gnutella, Gnutella2, and Direct Connect P2P file sharing protocols and in file sharing applications such as Phex, BearShare, LimeWire, Shareaza, DC++[4] and Valknut

Indeed.  I'm a poor programmer anyway, so I couldn't be of much help if someone wanted to implement this.  But I guess it wouldn't be very difficult since there is already a bitcoin JSON API.

I still think, however, that GNUnet is the only appropriate network.  Because as long as you pay someone to get something, there is the risk he doesn't give it to you.  This implies the need for an economics algorithm in the network, and such an algorithm is already implemented in GNUnet.

[mizerydearia]

I still think, however, that GNUnet is the only appropriate network.  Because as long as you pay someone to get something, there is the risk he doesn't give it to you.  This implies the need for an economics algorithm in the network, and such an algorithm is already implemented in GNUnet.

Good idea.  Perhaps you should consider contacting GNUnet dev team and present to them the same idea and maybe both dev teams can work together to establish something.

[grondilu]

I still think, however, that GNUnet is the only appropriate network.  Because as long as you pay someone to get something, there is the risk he doesn't give it to you.  This implies the need for an economics algorithm in the network, and such an algorithm is already implemented in GNUnet.

Good idea.  Perhaps you should consider contacting GNUnet dev team and present to them the same idea and maybe both dev teams can work together to establish something.

done.  I've sent a link to this thread on freenode #gnunet.

[grondilu]

An interesting aspect of information market is that it is, as far as I know, the only market where both transfers are, thanks to bitcoins, irrevocables and decentralized.

I mean : when you buy a piece of information, say a 32Ko block, you exchange pure information against bitcoin.  Both transfer doesn't imply the use of a trust party in any way, at least if you don't require compensation.  It's really a pure exchange model, from a technical point of view.

Therefore I think implementing a P2P information market software can be an interesting experimentation.

If I were to suggest a technique to do so, I would say that a P2P decentralized adjudication and compensation system would be installed, inspired from the bitcoin model.  Each bid and ask for a given hash would be broadcasted globally on the network, and a node would be allowed to perform both adjudication/compensation automatically after some proof of work of increasing difficulty.  The unpredictable location of this node would ensure honnesty of adjudication/compensation.  At least as long as the honnests nodes are majoritary, like in the bitcoin model.

Let me give an example.

Say Alice is a node in the network, and wants to buy the 32Ko block whose hash is HHH.  She lets every nodes in the network know what she wants (she broadcasts a bid).   Bob has HHH, and in response, he broadcast a ask for this block.  After some time, all nodes will have their own copies of a global orders book stored in their memory.  According to the protocol, nothing happens until a node, say the one that belongs to Claire, performs a certain proof-of-work on this book order.  Then Claire does the adjudication of the book order, and publishes her results, resulting in the possibility of the exchange between Alice and Bob.  Alice then send bitcoins to Claire, and Bob sends the block to Claire.  Claire, then, sends block to Alice, and bitcoins to Bob, thus performing compensation.

Anyway, I think the information market problem is a interesting one, from a theoretical point of view.

There would be many ways to implement it, imo.  Only one would be perfect, and we should keep looking for it.  For instance, the above proposal would suffer several problems.  One is that nothing is done to prevent dumb nodes to have the whole network waste time.  Indeed, a node might very well *try* to steal some bitcoins by pretending it has pretty much all hashes, although it actually has nothing.  The network would realize the scam and avoid it only at compensation level, thus wasting a lot of resource and time.  It could halt the whole network.  Thus, a trust system should be implemented, in order to avoid nodes that try to publish rubbish.  The economics algorithm of GNUnet would do the job, I guess.

An other idea :  in a "do one thing, but do it well" philosophy, adjudication and compensation should be separated.  In our example, this would mean that Claire would only be in charge of adjudication.  She would process the order book and produce a "request for compensation".  Such a request would then be broadcasted in the network, and an other proof-of-work would be required on it before the compensation could be performed by an other node, say Daniel.

Both Claire and Daniel, seen respectively as a Adjudicator and a compensator, would be judged in the trust GNUnet model by Alice and Bob for their respectives roles, so that they can be avoided later if they don't do their job properly.

[Timo Y]

Your model of paying for divulgation doesn't solve the free rider problem.

Still, lighthouses can be run profitably in a voluntary market so there is no reason that this can't.

Only difference is that it's likely to be less profitable than the paying for diffusion model (back when it used to work). You will never be able to completely eliminate free riders.

Another paradigm altogether is that instead of selling static data, you sell algorithms or services. Or rather, you sell their data output and never divulge the actual algorithm.

Computer game makers have already caught on to this. This is why you don't see that industry fighting to maintain the copyright staus quo the way the MAFIAA is doing.

[grondilu]

Your model of paying for divulgation doesn't solve the free rider problem.

Still, lighthouses can be run profitably in a voluntary market so there is no reason that this can't.

Only difference is that it's likely to be less profitable than the paying for diffusion model (back when it used to work). You will never be able to completely eliminate free riders.

I'm not sure that free riders are an issue as long as the seller can make sure he raised the expected amount of money.

If you buy a car, and then lend it to a friend, I'm not sure your friend can be considered as a "free rider".  Once your car is bought, you can share it with someone else : it's not the initial seller's business anymore.  Selling divulgation allows alienation, while selling diffusion doesn't.

Indeed, it's likely to be less profitable, of course.  But since sellers presently use public force to sell, I can understand that they make more money than if they wouldn't.  A gangster certainly makes less money with no gun.

[grondilu]

Very simple :   the beginning of the file foobar.avi directly available via http.

With a CGI script, clients could pay bitcoins to add some additionnal data to the file, until it is complete.  The price would be expressed in BTC/seconds.

Sounds so simple that something tells me it will never work.

[romsek]

IMHO for this hypothetic information market to be attractive it needs several pre-conditions like an existing anonymous p2p network where content is spread anyway. Else there won't be a big demand for such an information market: Without an anonymous network it's still more lucrative for an information distributor to sell information the traditional way. This may not apply to content where small amounts of earnings are expected.
Also necessary would be a rethinking of people about how they buy content and about the idea of property.

From the point of view of a consumer it will look like this:

• Buy block(s) for a quite low price if demand is high
• Wait for the expected money to be raised (if it will be raised at all!)
• When raised, download remaining blocks and finally be able to consume content
• Live with the fact that others will be able to consume content without ever having to buy blocks

From the point of view of an information seller

• Advantage that there is no need to provide a distribution infrastructure itself like file servers etc.
• Necessity to set a price one time without room to make more profit in the future with the same content

Will a purchaser risk to invest any money even if there is a chance that the expected money will not be raised at all?
Are people patient enough to wait for the complete blocks to be sold until they can consume the content? How long will it take? The less popular, the longer it will take?

[FreeMoney]

Your model of paying for divulgation doesn't solve the free rider problem.

Still, lighthouses can be run profitably in a voluntary market so there is no reason that this can't.

Only difference is that it's likely to be less profitable than the paying for diffusion model (back when it used to work). You will never be able to completely eliminate free riders.

Another paradigm altogether is that instead of selling static data, you sell algorithms or services. Or rather, you sell their data output and never divulge the actual algorithm.

Computer game makers have already caught on to this. This is why you don't see that industry fighting to maintain the copyright staus quo the way the MAFIAA is doing.

It will be really really sad if we ever completely eliminate freeriders. Hate against freeriders is hate against humanity. Every day I see and hear things without being hurt or threatened, landscaping, murals, some guy playing drums in his house every day when I walk to the beach, a movie at my friends house, thousands of awesome posts on forums. I am a freerider and I am not a problem.

[FreeMoney]

Necessity to set a price one time without room to make more profit in the future with the same content

Oh my god, imagine if hamburgers were sold that way!

[grondilu]

It will be really really sad if we ever completely eliminate freeriders. Hate against freeriders is hate against humanity. Every day I see and hear things without being hurt or threatened, landscaping, murals, some guy playing drums in his house every day when I walk to the beach, a movie at my friends house, thousands of awesome posts on forums. I am a freerider and I am not a problem.

Nice.
+1

[grondilu]

IMHO for this hypothetic information market to be attractive it needs several pre-conditions like an existing anonymous p2p network where content is spread anyway. Else there won't be a big demand for such an information market: Without an anonymous network it's still more lucrative for an information distributor to sell information the traditional way.

Of course.  But this first pre-condition is satisfied already :  P2P anonymous network exist.

Will a purchaser risk to invest any money even if there is a chance that the expected money will not be raised at all?

Setting a selling price and sticking to it is a business model in *a lot* of other economic sectors.  I don't see why it wouldn't work for information.

Are people patient enough to wait for the complete blocks to be sold until they can consume the content? How long will it take? The less popular, the longer it will take?

I don't know.  The thing is :  they don't have much choice.  It's this or they don't see content at all.  That's the point about ownership :  you're supposed to have some control on access to the thing you own.  Selling is about requesting money for this access.  If you don't want to pay, then you just have to forget about accessing to the thing you wanted.  It's a choice the buyer has to make.

[romsek]

Oh my god, imagine if hamburgers were sold that way!

But in grondilu's information market you can't sell your information hamburger more than one single time. With real hamburgers, you can.

Thanks for your reply, grondilu.

Setting a selling price and sticking to it is a business model in *a lot* of other economic sectors.  I don't see why it wouldn't work for information.

But would you buy a block for 5 $ if there is a chance that you will never get the complete content?

[grondilu]

But would you buy a block for 5 $ if there is a chance that you will never get the complete content?

Yes, as long as my desire to get the content overcomes my fear of losing 5$.

Of course, it's up to the seller to evaluate this probability and adjust his price.  Sometimes, he just won't find any acceptable price and will give up the idea of making the product.  It's not much different from conventionnal businesses.