My assumption regarding CoinJoin is its not centralized: its simply the idea that you can mix together a few transactions as one to make it unclear who is paying who. Doing this in a decentralized manner is nontrivial but possible and I assume what is intended here.

If you are interesting in another attempt to explain a few things, you can check out my brand new Skycoin FAQ on my new (unofficial) Skycoin wiki: http://skycoin.wikidot.com/faq

Eventually the content will likely be moved to some Aether based wiki, but for now the content can live there. Feel free to add questions and/or answers everyone. I make no guarantees about the quality of the content.

I see you merged my pull request and responded to my questions. Thanks! I haven't gotten a chance to go through your responses in detail yet, but I have setup a wiki since you expressed interest. http://skycoin.wikidot.com/

Due to details of the terms of service, that site will have to be for informational, not promotional purposes ("any material that promotes financial products" violates the TOS for wikidot). Meaning skycoin promotional content will have to live somewhere else (we likely don't want it on a publicly editable wiki anyway).


I can quickly/easily add some discussion forums there, as well as start collecting FAQ content and links to various other resources.

Should I setup some discussion forums? (I don't want to try and suck your followers over to my forums unless you want me to and you will participate there)

I'm still trying to get caught up on all the new stuff (I haven't made it through all the white papers yet), so just direct me to the right part of them if these points are addressed there.


There has been a big emphasis in your posts here on getting community involved in the writing and documentation over the last couple months. I see no wiki. Is there any action on this front? Are you waiting until you can build one Aether before starting this effort? I assumed the Whitepapers up on githhub were an attempt at this, but I don't see any calls for editors, and my pull request for some edits as just sat there for ~1 week now. I figured I might as well contribute edits as I read the docs, but if you aren't ready for that I'll read+edit them later. I'd be happy to setup a wiki if you want one and give you admin rights on it. I can't offer much of my time to moderate or administrate it though.


Proof of stake elections and 51% attacks: If you want to resist 51% attacks, is it really a good idea to give a party with a majority of the coins the power to do what ever they want with the source code? Would your mentioned plan of putting the source in the blockchain be used to push out auto-installed patches? (Developer elections -> put horrible malware in code -> store source in blockchain -> everyone update to malware). Any automation of patches is asking for someone to send out a steal all the private keys patch, but it could be a bot net patch aswell.... With deterministic wallets, you are pretty screwed if something like that slips in (most users will just have one key you need to steal, and then legit and attacker transactions are indistinguishable).

From a security perspective, it is your view that having exactly 1 implementation of skycoin is the best option, correct? (Just a clarifying your view here)

Given that you are looking at running downloaded code securely (at least I think thats what your application stuff is for), have you looked at http://genode.org/? They do nice things like per processes virtual file systems, have a secure microkernel etc. Run it as an is or vm/application on your host OS, and run your services inside it and you get robust isolation of native code. They have some done some nice work making their ports identified by a hash of their inputs lately (http://nixos.org/ style): it seems like a platform that might interest you.

The meshnet is intended to be a nice privacy tool with benefits comparable to tor, but lower latency correct?

The meshnet is intended to allow funding nodes via micropayments in skycoin to cover bandwidth costs correct? Doesn't this leave force all node operators to record detailed and published logs (on their personal block chains) describing all the transactions which inherently correspond to everyone who send data through their node? This seems like it would allow any third party to do traffic correlation attacks much like the ones on tor, except you don't need access to the connections. Even if they don't end up being publicly inspectable, logging everything seems like it might have some real issues (it can be requested by law enforcement, and takes up a lot of space)

The initial version is going to ship with centralized route finding server correct? This means if you want to connect to someone, you have to tell a third party about it, correct? It seems like this is not a Tor like privacy service until that's fixed. Is there reason to believe you will find a solution to this soon (or ever: its hard)?

How do you find a route to this trusted third party which will do route finding for you? I assume you will just special case it (don't use sender side route selection), but I'm curious if you have another design.

Most of the time I check out the master branch it is horribly broken. With the impending IPO, are you going to have stable release branches, which will generally work and have few (easy to review) changes? I have no interest in blindly running binary someone ships, or being stuck with some random old revision that happens to be known good and miss any fixes.

Can you guys please provide better commit messages? Looking at https://github.com/skycoin/skycoin/commits/master shows lots of commits with the same message just referring simple to something that I don't know what it is, some project area, of even just "changes", or "test". The last 3 were better: please continue that trend.

I expect a lot of nodes (say my phone) will want to simply ask a bunch of random nodes for the network state (block chain, and consensus status). It doesn't seem like there is an incentive for nodes to spend the bandwidth to inform anyone who asks of their consensus state, upload blocks etc. Are you just assuming enough people will bother to run nodes that provide these services that things will work? (It seems like a reasonable assumption, bitcoin nodes do exactly this, I just wanted to check). I suppose such requests could come prepaid for return bandwidth, though that just makes nasty lazy nodes that don't respond make money...


Is this the right place to post such questions? Should I just pose such things here, or is there a place not buried on 60+pages of overlapping discussions for addressing such topics? For stuff thats more issue like than question like I can post issues to github, but my issue there hasn't been responded to: https://github.com/skycoin/whitepapers/issues. Should I post such issues here as well?

Are you using cryptographic accumulators at all? I have to wonder if there is some neat cryptographic accumulator trick that could accumulate signatures.

To check that a given claim some node made in the past (some signed block from their personal block chain) is in the block chain they are currently publishing, you need to ask for the current node then get (at least the headers) of all the nodes back to the node you are interested in so you can check the hashes, correct? It seems you may have thought of this, but in-case you haven't, you know you could put the hash of the node 100 nodes back, and one 10000 nodes back etc. in each node (or just every 100th and 10000th etc respectively) and effectively get a skip list for fast checks. It seems like minimizing the cost of checking this kind of thing will be pretty important as part of making random audits of nodes cheap and easy.

Anyway, so far the consensus process using the personal block chains seems like a pretty robust/secure design. I'm looking forward to reading further into the whitepapers to see the details: getting it to be efficient storage, bandwidth and computation wise seems like an interesting challenge.

Is that supposed to be bad?

I'd rather evaluate someone for their works than their age. Aaron Swartz was doing great work at 14. If Skycoin is a 12-14 year old, that says little more than potential to improve. It would make much more sense to criticize Skycoins slipping deadlines, or some other actual issue with the design or project than the age: thats just counterproductive (and without evidence). And really, being a bronnie wouldn't effect anything; it's not relevant.

And now onto the real issues:

So if I understand correctly, Skycoin is basically providing (in addition to a currency) tools for building a distributed reputation system. A personal block chain is a set of claims about the identity associated with your public key, and you provide a system to replicate and inspect it. The claims may have multiple signers, and the block chains of those signers may also be inspected.

This basically provides (in a distributed manner) one of the key features of Silk Road: its a way to build reputations associated with identities which you can send messages and money to.

However, users of the system (developers of software that uses these features) still have to sort out all the nasty issues: if you can make an identity with a block chain, you can make a billion of them. If trust can be built through relations, your billion clones will do very well. All of them can edit their block chains at well, and all agree with each-other while editing them, or throw them away and be replaced. I could instantly create an exchange that appears to have a long history of millions of transactions with millions of users (my clones). The moment it gets some money sent in, it can send it off into the pool, and create a new exchange with a similar history of transactions between other identities I control. You still need a robust web of trust to prevent this, and thats a hard (and non user friendly) problem to solve.

Its a nice service to expose, but I don't see any particular benefits it offers, or great uses without cross chain transactions, which likely are impossible in such a case. If you could do bonded transactions (requires cross chain transactions), of have a system for assigning trust (like a clone of Ripple, thats not so amazing) it would be more interesting.

Leveraging proof of stake by putting proof of sake in SkyCoin in your personal block chain (or signing your block chanin and putting that signature in the SkyCoin block chain) might be useful here. However, the same skycoin can can be mixed and reused to sign your next fake identity (or next million of them), so I don't see how that helps at all, unless you rely on using coins that haven't been transferred or used for trust in a long time. That however destroys fungibility and encourages handing over private keys instead of creating transactions (the value of non transferred coins would be higher): that would wreck the currency.

I just don't see whats so great about it: it doesn't solve distribution of trust or sybil attacks as far as I can tell.

I should note that I'm not saying such a system is not totally useless: if you already have a system for assigning trust, it would be a perfectly fine setup for replacing our existing SSL certificate mess for example (Each authority and sub authority gets a block chain, replicated as needed). Nothing very new there, but the libraries could be useful for that kind of thing.

Personally I find NxtCoin's Colored Coins http://www.nxtcrypto.org/nxt-coin/nxt-colored-coins more powerful since they are tied to the main block chain. You can have your own personal info associated with them, but you can also involve them with transactions trustlessly since they are part of the main block chain (no need for cross chain transactions). Associate those with contracts and you get a system that still doesn't solve the entirety of the exchange problem, solves more of it than personal block chains do.

I could create my own currency/block chain with NxtColored coins. I set their value to be 1 btc. I could then securely trade my colored coins for how ever many Nxt a Btc is worth at the time. I'd still need to go off chain to give real Btc for one, but it could be securely traded for someone else's colored coins that represent Btc, or Doge for example. Its not a complete solution (no cross chain transactions with existing currencies), but it effectively allows secure cross chain between all Nxt bases currencies / colored coins. Thats pretty useful, and not possible with personal block chains.


In short: I don't think all the grand claims about features possible with personal block chains are that great: personal block chains might be a fun toy for devs, but they only solve a small part of the problem (part of the tooling). Nxt offers more powerful features that do more to help implement these kinds of things. Skycoin could support colored coins though.

A modular design that allows easy code reuse for things like personal block chains is great, but I don't think the grand claims about them are warranted: focus on the core feature, the consensus algorithm. Grand claims about that are more interesting and details would be even better.

It seem like most of these proposed features (enforceable contracts, distributed trust-less exchanges, reputation systems etc) all come from the core feature of cross chain transactions. Cross chain transactions allow adding other currencies/goods to the economy that can be securely exchanged. This is a very powerful feature which solves tons of problems including the scalability ones (you can shard the currency if the distributed consensus becomes too expensive, or even add ultra low transaction cost centralized sub-currencies that can be converted back to any other currency)

I've looked into cross chain transactions before: the idea is not at all new. I've just never seen anyone claiming they had a way to do it (If I missed one, please direct me to it. ZeroCoin seemed kinda close, but didn't solve the general problem as far as I could tell). I tried to design some solutions myself, but I was unsuccessful (No surprise since I'm not an expert). If Skycoin really does solve the cross chain transaction problem, then I think we really have something here. I hope this is the case.

However I have a couple of doubts: Skycoin appears to claim supporting cross chain transactions involving existing currencies that have their own block chains already that are not effected by skycoin's. This means that actions on the skycoin side (such as a transaction failing to get included due to being involved in a double spend) can not effect the other block chain. If one version of a skycoind/bitcoin trade transaction gets accepted on the bitcoin side, but a double spend is issued on the skycoin side, I don't see how both sides would be guaranteed to reach the same consensus. Even more confusing to deal with: suppose bitcoin forked, what happens then? A block chain forking is a perfectly valid and acceptable thing (if there is a legitimate disagreement about some transaction's validity, then its forked. Usually one side is mostly abandoned (but still existent), but thats not guaranteed)

Secondly: I don't see how these problems could be solved, even within skycoin one block chains. Solving it there seems more likely to be possible, but I still don't see how it can be done. I won't say it can't be done (cryptography does pretty amazing things), but the explanation for how that works is the main thing I'm looking for.

To be clear though, even if Skycoin fails to deliver a cross chain transaction solution, it still looks like it has a lot of potential, but I'd start to be more suspicious if such a significant claim ended up being false.

I had the same issue. I assume its related to the bug I filed here: https://github.com/skycoin/skycoin/issues/16

Basically the current version  ( 5afc6d2105b29c59e2259d1a842ad6f6458417ec ) doesn't seem to work, and there hasn't been a commit in the last ~6 days, so its just been in this broken state for a while. As least thats how it seems to me.

Even if I had a fixed version, I couldn't distribute it due to licencing issues (So I'm not going to spend time looking into it further): I still haven't found where the license is specified so I can comply with the terms if distributing. Knowing it's "Its either GNU or MIT" is nice, but I can't comply with even the MIT terms without knowing the copyright... (Note, GNU isn't even a license, its just not unix. GNU GPL version 2 is a license (so is GNU GPL v3 or GNU LGPL). And that seems like a rather silly duel license choice anyway)

Apparently there is a license for it on github somewhere: has anyone found it? I looked through the repo, readme and wiki, and searched the github pages: nothing to be found.

Since a few people seem to be having issues getting SkyCoin running, I'll say what I did. Note: I foolishly didn't take notes while doing this, so something might be missing. I tried using the "setup.sh" script, but it messed with my .bashrc and seemed to need gvm which it apparently failed to install. I choose not to use gvm since it seemed needless and messy, and not to mess with linking directories around: I just added another GOPATH entry.

Setup guide:
Since I haven't reviewed all the code, I'm not going to run in on a computer I care about. Thus I setup a virtual machine (I choose linux mint in VirtualBox). I take no responsibility for what all this does to your computer and/or network. Take the proper precautions!

I installed (via sudo apt-get install) git, mercurial, libgmp-dev, libc-dev. I also installed Go 1.2.1 from here: https://code.google.com/p/go/downloads/list (You need at least 1.2, and my package manager only had 1.1)
I setup a GOPATH environment variable (http://golang.org/doc/code.html#GOPATH) in my .bashrc with 2 paths (colon separated), one for third party libs (the first entry on GOPATH where Go get puts stuff be default), and one for the skycoin repo. You can put them both in one if you want.

If you install Go in a non standard location, you need to set $GOROOT.

Add go to the $PATH in .bashrc: "export PATH=$PATH:$GOROOT/bin"

create the path: $GOPATH/src/github.com/skycoin/
and in there "git clone https://github.com/skycoin/skycoin.git"

This gives you a path "$GOPATH/src/github.com/skycoin/skycoin" (Note the skycoin/skycoin, you must have both directories)

In there, "go get -u" to download/update all dependencies.

"./run.sh -web-interface=true" will get you a web server you can view at http://127.0.0.1:6402 for the UI.

Currently if I run the UI it keeps printing panics, and does not seem to be working, but at least it runs shows a UI and prints lots of things.



If you follow this, please let me know if it works or need and changes/improvements.

I assume that's a question about mining. There is no mining in SkyCoin. Its proof of stake based, not proof of work based. If I understand correctly, that hash function is used to generate addresses in a deterministic wallet: its not part of the coin protocol itself, nor do you have any reason to care about its performance unless you are trying to generate lots of addresses for deterministic wallets.

What license is the code under? Currently I don't see any license on the client or the darknet project. That implicitly puts it under standard copyright, which is pretty crippling; I won't use a client I can't fork and patch.