4 ms·

Hey, Chris Anderson here, this article is about my project. Here are slides from my Open Source Bridge talk on the subject. First half is philosophy the second

by jchrisa 12y ago

Hey, Chris Anderson here, this article is about my project. Here are slides from my Open Source Bridge talk on the subject. First half is philosophy the second half is tech: https://speakerdeck.com/jchris/subjective-crypto-currency https://speakerdeck.com/jchris/subjective-crypto-currency

rdl 12y ago
Perhaps you should rename it "whuffie".
- chadgeidel 12y ago
  This is the first thing I thought of. For those that haven't read the book: http://en.wikipedia.org/wiki/Down_and_Out_in_the_Magic_Kingdom http://en.wikipedia.org/wiki/Down_and_Out_in_the_Magic_Kingd...
- jchrisa 12y ago
  Down and Out is a good one, I also like http://www.shareable.net/blog/the-guy-who-worked-for-money http://www.shareable.net/blog/the-guy-who-worked-for-money I think there is something problematic about those stories, mostly because they insist on making reputation into a single scalar value.
  - blazespin 12y ago
    Agreed, your improvement is important.
zz1 12y ago
Looked at the slides, I kind of undestand something now, and it looks intriguing if not really interesting. Have you thought about key revocation, how is it handled?
- jchrisa 12y ago
  No it's hard and could lead to sad stories. Say your phone holds a few wallets and a hacker breaks in a starts double spending. That would ruin those wallets and I guess there are simpler cases like you just lost your wallet / forgot your password that might be as troubling. So key handling is totally one of those things I'd love to outsource. Actually please let me outsource as much of of the crypto as possible, I don't want to be inventing things. To that end http://www.w3.org/TR/WebCryptoAPI/ http://www.w3.org/TR/WebCryptoAPI/ looks like a reasonable start. I've also been reading http://datatracker.ietf.org/doc/draft-ietf-jose-cookbook/ http://datatracker.ietf.org/doc/draft-ietf-jose-cookbook/ and generally trying to get a sense of the least opinionated path I can take. Luckily our iOS lead at Couchbase is experienced writing apps that use cryptographic signing, so he's written up the beginnings of a spec that we'll likely evolve: https://github.com/couchbase/couchbase-lite-ios/wiki/Signed-Documents https://github.com/couchbase/couchbase-lite-ios/wiki/Signed-... If you want to join in the implementation nitty gritty feel free to start a thread in our group: https://groups.google.com/forum/#!forum/mobile-couchbase https://groups.google.com/forum/#!forum/mobile-couchbase I've always used my own application development goals to find interesting directions at Couchbase so this project is just an example of how we work.
  - MarkPNeyer 12y ago
    my friend and i have been working on something like this, which may have the answer to the key revocation problem. in our system (which we are calling dewdrop - https://github.com/neyer/dewdrop https://github.com/neyer/dewdrop - still very much a work in progress) you don't pass coins around; you make statements. you can make statements of the form 'i trust person X', and then statements made by person X are given more weight for you, as are statements by those X trusts, etc. in our system, key revocation would be just another statement. if i say "hey this is mark, my phone was captured at this time and these N transactions are invalid", my friends - if they believe me - can say 'i believe he is telling the truth here.' everyone who trusts us will then ignore statements made from that address during that time period. the statements will still be on the network, but people who trust me and my friends will not believe i made them. because there is no transfer of coins - just statements that anyone can make - there's no issue of 'do we really need to revoke the key?' people who trust me, and who trust my friends, will see those statements as having been revoked; people who don't trust me will need to decide for themselves on this.
anigbrowl 12y ago
Thanks for chiming in. I find it easy to conveice of a multidimensional reputation, eg your HN karma encapsulates one vector, the number of upvotes on Reddit or some special-interest forum another, and so forth. The bit I don't fully understand right now is how you approach convertibility. Say I'm into motorcycles, and I am a regular at motorcycle-forums.com where I dispense helpful advice on everything from tire pressure to tappets. Over time I deservedly acquire a reputation as That Helpful Guy, and the grateful votes of other users accumulate. One day I decide I want a new bike, and head over to the marketplace subforum to purchase one. Turns out that Alica has just the bike I want! Naturally I want to use my accrued Biker Points to get a discount. Now, in a closed economy. Alice might be delighted to have my BP - she could trade them for other goods and services with other bikers, and there might even be a social value in having dealt with me (a good few specialist websites maintain informal 'trusted buyer/seller' lists which saves members money vs. low-trust high-fee markets like eBay). But what if Alice is done with biking and just wants to get rid of her Vroomatic 5000 so she can take up windsurfing or something? My BP will be of little use to her on the ocean wave, and there may not be enough transitional activity between the Biker and surfer communitites to establish a maingful rate of exchange, so instead you end up with a bunch of shady pseudocurrencies, sort of like penny stocks. One way I can see it working is if quality websites charge a small joining or membership fee and a portion of that constitutes an initial pool of points. Alternatively, a portion of ad revenue might be convertible into points, as those members who consistently make valuable and informative contributions are a key part of the site's draw, so it makes sense for the operator to subsidize the most 'productive' members. I may have got completely the wrong end of the stick here, and you only meant to substitute reputation for cryptographic proof-of-work, but I'm guessing form your slides about inequality and different concepts of value that you're looking to examine the notion of wealth in more subjective contexts, rather than just authentication.
- jchrisa 12y ago
  Yes the notion is that we can throw away the idea that everyone must be ranked on a single number line. Instead each person would see their own view of who is wealthy, who is reputable, who is a hard worker, etc etc. And depending on my values, I might disagree with your view of a particular person. So the currency is meant to reflect that.
pessimizer 12y ago
This is almost exactly the same idea that I had - it may differ in specifics, but they map to the same thing, so I'd be interested in following the progress of its design. Have you thought of the possibility that, for example, universities could issue coins to their graduates, and scientists could issue them to papers they have reviewed (and the value, in that case, could well be the actual review)?
- jchrisa 12y ago
  Coins can really be any data, issued by anyone. So they can almost be thought of as contracts in some cases, assertions in another. When the video of my talk is up you'll see that comes up in the Q/A. Part of the reason I'm being late-binding about the technical implementation is because I know I haven't come up with anything exactly new. I've been working a long time to solve the database sync problem, and now that we shipped, I'm thinking about what we can build on top of it. Which particular key scheme we use is of secondary importance, the important thing for me is exposing crypto primitives as part of a distributed database.
kaoD 12y ago
From the article and the slides, I imagine these as p2p federal tradeable karma points. Did I get the big picture?
facepalm 12y ago
I am a bit floored because it sounds like a thing I have been thinking about (floored because with your credentials, you are much more likely to succeed than I am :-) ). I would have called it LikeCoin. My starting point is that likes have some value. A baker might give you some bread in exchange for a Facebook like. However, that would only work once or twice - the more you spend it, the more it loses in value. On the other hand, the more followers you have, the more valuable your likes are. I just thought that could be modeled into a coin without the dreaded "double spending" risk - double spending would be no problem, because everybody could create as many "coins" as they want anyway. But they would have to worry about inflation if they create too many coins...

by jchrisa 12y ago

rdl 12y ago

Perhaps you should rename it "whuffie".

chadgeidel 12y ago
This is the first thing I thought of. For those that haven't read the book: http://en.wikipedia.org/wiki/Down_and_Out_in_the_Magic_Kingdom http://en.wikipedia.org/wiki/Down_and_Out_in_the_Magic_Kingd...
jchrisa 12y ago
Down and Out is a good one, I also like http://www.shareable.net/blog/the-guy-who-worked-for-money http://www.shareable.net/blog/the-guy-who-worked-for-money I think there is something problematic about those stories, mostly because they insist on making reputation into a single scalar value.
- blazespin 12y ago
  Agreed, your improvement is important.

zz1 12y ago

Looked at the slides, I kind of undestand something now, and it looks intriguing if not really interesting. Have you thought about key revocation, how is it handled?

jchrisa 12y ago
No it's hard and could lead to sad stories. Say your phone holds a few wallets and a hacker breaks in a starts double spending. That would ruin those wallets and I guess there are simpler cases like you just lost your wallet / forgot your password that might be as troubling. So key handling is totally one of those things I'd love to outsource. Actually please let me outsource as much of of the crypto as possible, I don't want to be inventing things. To that end http://www.w3.org/TR/WebCryptoAPI/ http://www.w3.org/TR/WebCryptoAPI/ looks like a reasonable start. I've also been reading http://datatracker.ietf.org/doc/draft-ietf-jose-cookbook/ http://datatracker.ietf.org/doc/draft-ietf-jose-cookbook/ and generally trying to get a sense of the least opinionated path I can take. Luckily our iOS lead at Couchbase is experienced writing apps that use cryptographic signing, so he's written up the beginnings of a spec that we'll likely evolve: https://github.com/couchbase/couchbase-lite-ios/wiki/Signed-Documents https://github.com/couchbase/couchbase-lite-ios/wiki/Signed-... If you want to join in the implementation nitty gritty feel free to start a thread in our group: https://groups.google.com/forum/#!forum/mobile-couchbase https://groups.google.com/forum/#!forum/mobile-couchbase I've always used my own application development goals to find interesting directions at Couchbase so this project is just an example of how we work.
- MarkPNeyer 12y ago
  my friend and i have been working on something like this, which may have the answer to the key revocation problem. in our system (which we are calling dewdrop - https://github.com/neyer/dewdrop https://github.com/neyer/dewdrop - still very much a work in progress) you don't pass coins around; you make statements. you can make statements of the form 'i trust person X', and then statements made by person X are given more weight for you, as are statements by those X trusts, etc. in our system, key revocation would be just another statement. if i say "hey this is mark, my phone was captured at this time and these N transactions are invalid", my friends - if they believe me - can say 'i believe he is telling the truth here.' everyone who trusts us will then ignore statements made from that address during that time period. the statements will still be on the network, but people who trust me and my friends will not believe i made them. because there is no transfer of coins - just statements that anyone can make - there's no issue of 'do we really need to revoke the key?' people who trust me, and who trust my friends, will see those statements as having been revoked; people who don't trust me will need to decide for themselves on this.

anigbrowl 12y ago

Thanks for chiming in. I find it easy to conveice of a multidimensional reputation, eg your HN karma encapsulates one vector, the number of upvotes on Reddit or some special-interest forum another, and so forth. The bit I don't fully understand right now is how you approach convertibility. Say I'm into motorcycles, and I am a regular at motorcycle-forums.com where I dispense helpful advice on everything from tire pressure to tappets. Over time I deservedly acquire a reputation as That Helpful Guy, and the grateful votes of other users accumulate. One day I decide I want a new bike, and head over to the marketplace subforum to purchase one. Turns out that Alica has just the bike I want! Naturally I want to use my accrued Biker Points to get a discount. Now, in a closed economy. Alice might be delighted to have my BP - she could trade them for other goods and services with other bikers, and there might even be a social value in having dealt with me (a good few specialist websites maintain informal 'trusted buyer/seller' lists which saves members money vs. low-trust high-fee markets like eBay). But what if Alice is done with biking and just wants to get rid of her Vroomatic 5000 so she can take up windsurfing or something? My BP will be of little use to her on the ocean wave, and there may not be enough transitional activity between the Biker and surfer communitites to establish a maingful rate of exchange, so instead you end up with a bunch of shady pseudocurrencies, sort of like penny stocks. One way I can see it working is if quality websites charge a small joining or membership fee and a portion of that constitutes an initial pool of points. Alternatively, a portion of ad revenue might be convertible into points, as those members who consistently make valuable and informative contributions are a key part of the site's draw, so it makes sense for the operator to subsidize the most 'productive' members. I may have got completely the wrong end of the stick here, and you only meant to substitute reputation for cryptographic proof-of-work, but I'm guessing form your slides about inequality and different concepts of value that you're looking to examine the notion of wealth in more subjective contexts, rather than just authentication.

jchrisa 12y ago
Yes the notion is that we can throw away the idea that everyone must be ranked on a single number line. Instead each person would see their own view of who is wealthy, who is reputable, who is a hard worker, etc etc. And depending on my values, I might disagree with your view of a particular person. So the currency is meant to reflect that.

pessimizer 12y ago

This is almost exactly the same idea that I had - it may differ in specifics, but they map to the same thing, so I'd be interested in following the progress of its design. Have you thought of the possibility that, for example, universities could issue coins to their graduates, and scientists could issue them to papers they have reviewed (and the value, in that case, could well be the actual review)?

jchrisa 12y ago
Coins can really be any data, issued by anyone. So they can almost be thought of as contracts in some cases, assertions in another. When the video of my talk is up you'll see that comes up in the Q/A. Part of the reason I'm being late-binding about the technical implementation is because I know I haven't come up with anything exactly new. I've been working a long time to solve the database sync problem, and now that we shipped, I'm thinking about what we can build on top of it. Which particular key scheme we use is of secondary importance, the important thing for me is exposing crypto primitives as part of a distributed database.

kaoD 12y ago

From the article and the slides, I imagine these as p2p federal tradeable karma points. Did I get the big picture?

facepalm 12y ago

I am a bit floored because it sounds like a thing I have been thinking about (floored because with your credentials, you are much more likely to succeed than I am :-) ). I would have called it LikeCoin. My starting point is that likes have some value. A baker might give you some bread in exchange for a Facebook like. However, that would only work once or twice - the more you spend it, the more it loses in value. On the other hand, the more followers you have, the more valuable your likes are. I just thought that could be modeled into a coin without the dreaded "double spending" risk - double spending would be no problem, because everybody could create as many "coins" as they want anyway. But they would have to worry about inflation if they create too many coins...