Silk Road forums

Discussion => Silk Road discussion => Topic started by: joywind on October 04, 2013, 02:42 am

Title: Successor to Silk Road
Post by: joywind on October 04, 2013, 02:42 am
This is I believe the best approach to a fully decentralized anonymous drug marketplace.

It seems to me that SR essentially serves four functions:

* A feedback rating system to allow users to select reliable vendors.
* A listing service allowing vendors to advertise their products and prices in a standardized format, and buyers to easily compare.
* A messaging service that allows buyers and sellers to contact each other and coordinate a transaction.
* A payment intermediation system that holds payments in escrow until delivery and arbitrates disputes.

First the escrow system probably has to go in a decentralized model. It's basically unnecessary as long as the feedback system is reliably trustworthy. Also it is by far the most illegal since it touches money laundering and all kinds of nasty stuff. Probably the most traceable too. And without a central authority who runs the escrow accounts? So I'd get rid of that. But you could debate this point.

It would seem to me that the remaining three functions should be separated into distinct services. One it severely reduces legal liability. For example if I run a feedback service I'm not even aware of what people are selling or buying, I simply accumulate a point ranking. Similarly a message service could be used for anything.

Second it reduces the systematic failure component. The most legally targeted is probably the lister. But if he gets taken down, he's also the most replaceable in the system. If the feedback and messaging services are still in place users can simply point to another listing service pretty seamlessly.

The listing system is essentially just a distributed hash table. Same as Bittorrent or Freenet. Just point to one or multiple trackers to access product listings. Anyone who wants to list would simply add a listing to the DHT in a structured manner (product, price, quantity, description, messaging info). The buyer simply inputs his parameters to find a list of relevant sellers. Decentralized hash tables are already solved and good libraries exist. Simply apply this to Tor in a market-based format.

Anonymous TOR and encrypted messaging is also not only solved, but several already good services already exist. TORChat seems pretty good, but the key to robustness is for most users to be using at least several options. No need even to do anything on this front.

Finally decentralized feedback is the least developed. What I envision here is a decentralized feedback protocol that can host multiple networks but using a single block chain. Multiple feedback networks live on top of the block chain, each one with their own independent ratings. A given feedback network accumulates the ratings for every user that subscribes to that network. A user can point to multiple feedback networks and each one will have a different feedback listing tied to a simple plaintext to a public key. Listings will be signed with the public key, so securely verifying the rating of the lister is easy.

The biggest problem is that feedback networks need to be kept honest, remove spam reviews, etc. How do you do this without a central authority? The first approach is to require that every feedback transaction burn a small bitcoin transaction fee. This makes it prohibitively expensive to do a bunch of sham transactions to inflate your feedback rating. The transaction fees will be recovered by the miners on the block chain.

Second Set up the protocol so that within any given network the top 10-50 users on any given feedback network can arbitrate feedback as good or spam. In effect decentralize the administrative authority to the highest rated users. They're incentivized to keep the network honest, because they hold a high rating on that network. If bad feedback floods the network it reduces the value of their earned reputation on it.

In the case of a network being compromised by bad feedback and taken over by the spam raters, users can simply point to another trusted feedback network and use those ratings instead. In fact users will probably use several distinct feedback network to increase the diversity of the reviews when evaluating sellers. Having a multitude of trusted feedback networks makes it easy to detect when one becomes corrupted by comparing against other trusted networks.

Throw it all together into a single client program that makes these three functions in a relatively simple to use interface and I think you'll have a pretty competitive offering.
Title: Re: Silkroad 2.0
Post by: joywind on October 04, 2013, 02:49 am
This looks promising--

https://github.com/Bit-Wasp/BitWasp

Quote
"BitWasp is an open source project which aims to lower the barrier for anyone to set up there own secure, anonymous marketplace. It is envisaged that BitWasp may be used on Tor hidden services. As such it is designed with speed in mind. This project will also eventually full integrate with BitCoin providing a secure escrow service for buyers and sellers. This project is still in its very early stages. There is a preliminary planning document at https://piratenpad.de/p/bitwasp-planning

"WARNING - BitWasp is NOT production ready

"This project is very much under development and we have not yet made a beta release. Please be aware that this project has not yet undergone extensive security testing and the code base is still missiing a number of key features. Please download and test the code by all means but we would strongly recommend it is not deployed on a production system. We will update this README when there is a beta release."

It is seemingly built around decentralization ala p2p/bittorrent, so once it started up it could live on without a leader. Also, if it was set up for no commissions/ego juicing. With no 'figure heard' it would take away the ability to prosecute someone like DPR (or inigo or whomever Ulbricht actually is).
Title: Re: Silkroad 2.0
Post by: This_is_not_SOCA on October 04, 2013, 03:00 am
Some good concepts there Joywind

DHT has got to be the way to go for listings - the escrow point I would tend to agree with if the feedback system works well. Increasing the 'distance' between the BTC transfer and the listing itself would certainly be beneficial and muddies the waters in a good way.

Mmmm.....
Title: Re: Silkroad 2.0
Post by: Gridlokk on October 04, 2013, 09:03 am
subbed
Title: Re: Silkroad 2.0 (but not Silkroad 2.0)
Post by: This_is_not_SOCA on October 06, 2013, 11:41 pm
So I been doing a lot of thinking and this is what goes into post number 100 (are we there already - midnight is fast approaching for this Cinderella)

I was thinking about how Joywind had broken out SR functions and treated them independently and actually I think that is a good way of looking at it as independence of layers is key to everything. I see two stages which will later translate to 2 layers of the market.

Stage (Layer) 1- 'We' build the peer to peer, distributed, decentralized, technical market infrastructure which works to fulfill the following basic functions in a resilient way:

a) User Database including both vendors and buyers - there isn't one. All 'users' prove authenticity using PGP only - i.e. the ability to sign a message or challenge. PGP ties everything together. No user management - no superusers. Perhaps we run our own hidden service keyservers or maybe just use the clearnet publlic ones - or maybe no keyservers at all - swings and roundabouts.
b) Listings - a distributed file system - sellers can upload signed listings in a specific XML type format
c) Transactions - buyer creates and signs order and encrypts to buyer, vendor acknowledge by signing - BTC transaction conducted out of band (no BTC stored in this layer at all) - buyer and seller updates and signs transaction - transactions also get recorded in the DFS
d) Feedback and reviews - combination of key signing (to indicate trust) and signed feedback messages added to the signed transaction already in the DFS

The above can work purely on a peer to peer basis within TOR as an overlay network implemented using hidden services. Augmenting the user nodes with a few dedicated VPS would no doubt improve the overall performance of the network. There is no fee or markup at this layer, no commission, nothing. Just true p2p infrastructure. it would be possible for a buyer and a seller to operate using this infrastructure alone although unless somebody wrote a nice pretty unified GUI it would be a far more 'technical' experience than Silk Road...

Stage (Layer) 2 - HOWEVER, On top of this decentralized platform, others would be free to come along and create web applications which act as a friendly interface to the decentralized infrastructure above. This would be more for the Silk Road experience from the users point of view. It would be up to the web application provider to deal with interfacing into the peer to peer decentralized infrastructure hiding this complexity from the user be they a buyer or a seller. The owners of the web application can brand their site whatever they like, they are free to offer other services such as escrow - advanced messaging, anything they can think of etc etc. They would of course add a mark up to the basic transaction - up to them. There is the opportunity here for site operators to make money although they are likely to be at increased risk as they bring everything together and offer 'dangerous' facilities such as escrow, mixing etc. They may even choose to run their own user database (with passwords or whatever) although buyers/sellers should all exist as 'users' in layer 1 (i.e. have PGP keys). A web site could even exist on clearnet and back onto the layer 1 infrastructure but it's operators and users would probably not last very long.

Having 2 layers breaks the link between the hidden service web sites (easier to use but also easier to take down) and the underlying market infrastructure which would now be completely decentralized and hopefully harder to take down and subvert.

Really we are talking about building an overlay file system (there are a number out there that could fit the bill although most are not designed to work in TOR as hidden services out of the box) and some standardized XML formats for the various types of message (listings, transactions, feedback etc).

There is certainly some work but by reusing what is currently out there and keeping the various components as independent as possible, the layer 1 infrastructure could be in place without hundreds of man hours of time.

The layer 2 would be what made the platform a success and could create some nice diversity. If one of those layer 2 sites gets shut down or ceases operating - we all still know who everyone is - everyone still has their market identity (their pgp key) and the infrastructure is still all running - just choose another layer 2 provider.

I like this model - I believe it can creates great strength in the anonymous market. Any thoughts, good bad or ugly I'd be interested to hear them.

Title: Re: Silkroad 2.0
Post by: The Godfather on October 07, 2013, 02:14 pm
Have a look at the progress we have made so far:
http://postimg.org/image/x8j3thcnh/
http://postimg.org/image/k7qd0cvxt/
Title: Re: Not Silkroad 2.0
Post by: This_is_not_SOCA on October 08, 2013, 12:48 am
I will try to document some of this during the week and will make it available here if here is still here - although not in this thread unless Joywind changes the title(!). Yes there are lots of unaddressed issues but many of the can be worked through - it seems that not too many people have been layering distributed file systems on TOR which is a shame  and that is a key area to understand better - Freenet could potentially offer the file system from the get-go but has it's own set of problems and constraints. Some interesting discussions did take place on the Dissent thread ( dkn255hz262ypmii.onion/index.php?topic=208976 ) last month.

Freenet<>TOR hidden service gateways are certainly feasible but then we end up back with a requirement for dedicated servers or market participants willing to run 24/7 TOR and Freenet to act as gateways which is a non-starter.

A purpose built DFS using TOR hidden service would be highly desirable - I may be missing a trick but nobody seems to have bothered with this or at least publishing it - probably because of potential for p2p file-sharing abuse and the fact the freenet/i2p are already out there and up and running.

Title: Re: Silkroad 2.0
Post by: MarcelKetman on October 08, 2013, 12:58 am
Interesting. Subbed.
Title: Re: Successor to Silk Road
Post by: This_is_not_SOCA on October 08, 2013, 01:24 am
Thank you Joywind :)