News for hackers. Stuff that matters.
Re:Blockchain's Numbers Don't Add Up by wslh on Thursday September 08, 2016 @07:55AM (#12452146)
The problem is when you try to extrapolate the blockchain concept to old financial institutions and you really mean "modernization". It is very difficult to read articles about blockchains from financial institutions, they are just forcing a concept and deceiving.

I have also a company in this sector. In our talks with banks, stock exchanges, and government they say they want a blockchain. When we ask how many nodes they will run, the answer is usually one.

I also believe the cryptocurrencies and blockchain concepts have a lot of applications in Internet, but currently they mainly offer solutions for the dark and lighter nets.

Re:Blockchain's Numbers Don't Add Up by zaphar on Thursday September 08, 2016 @09:34AM (#12452805)
That is because they don't want bitcoin. They want a cryptographically verifiable merkle tree. Which is fine, it solves a number of problems that they have. The problem is that they keep conflating it with bitcoin which does considerably more than just maintain a merkle tree.
Re:Blockchain's Numbers Don't Add Up by guylepage3 on Thursday September 08, 2016 @09:26AM (#12452736)
Sorry, what I was stating has nothing to do with finance at all. We've created a new web. A truly decentralized web.

The blockchain should only be used for non-financial applications.

Re:Blockchain's Numbers Don't Add Up by lucozade on Thursday September 08, 2016 @10:17AM (#12453193)
> The blockchain should only be used for non-financial applications

Do you really mean it shouldn't be used in any financial applications or just not to replace fiat currencies? If the former could you elaborate on why? I'd be very interested in your opinion.

Re:Blockchain's Numbers Don't Add Up by alanwatts on Thursday September 08, 2016 @10:59AM (#12453606)
Perhaps it is because the fundamental nature of "finance" as a social conception is inherently centralized.

So any purposefully decentralized/distributed system used for financial (centralized) purposes becomes corrupted by default.

Re:Blockchain's Numbers Don't Add Up by nix0n on Thursday September 08, 2016 @10:20AM (#12453224)
https://blockstack.org/docs/how-blockstack-works has some interesting info.

I'm on board with the idea of a DNS replacement using a distributed hash table but I'm not clear on why Bitcoin specifically or a blockchain generally need to be involved.

Re:Blockchain's Numbers Don't Add Up by jude- on Thursday September 08, 2016 @11:59AM (#12454204)
The problems with using DHTs for naming are:

* Locality-unaware routing. Your lookups can bounce around the planet a few times before reaching the right key, leading to extremely high and extremely variable latency. This can be mitigated with DSHTs (e.g. CoralCDN from NYU) and/or with some clever key/value distribution schemes (e.g. Beehive from Cornell).

* Route censorship attacks. DHTs are vulnerable to sybils, which can take over all the routes for a particular set of keys and censor them.

* DDoS attacks. Nothing stops you from inserting a bunch of junk key/value pairs and overwhelming the DHT nodes.

* Churn attacks. A few sybils can add a bunch of nodes, stay online for a while, and then disappear over a phased withdrawal period. This leading to high node churn, which leads to content unavailability and network partitions.

* Content attacks. Unless the DHT's key is the cryptographic hash of its value (not the case in DHT-based DNS), nothing stops you from inserting invalid routes for the same name. Signing the value with a keypair doesn't help either, since that simply changes the name-to-IP lookup problem into a name-to-key lookup problem (i.e. how do you know the "right" key produced the signature?).

* Route hijacking. Again, if your DHT isn't content-addressed, sybils can simply take over routes for a set of keys and serve you the wrong values intentionally.

The reason Blockstack uses a blockchain to bind a human-meaningful names to hashes is because blockchains are decentralized, inimitable, and totally ordered. Each blockchain peer has a 100% replica of the system state, so the above limitations do not apply. Since they're expensive to imitate and since they're totally ordered, they can be used to enforce name acquisition on a first-come first-serve basis.

But of course, the trade-offs are that writes are slow, writes require PoW, and space requirements grow linearly. Blockstack in particular addresses this by using a blockchain to bootstrap name/key bindings, and then storing signed data off-chain in commodity storage providers (peer-reviewed paper at https://blockstack.org/blockstack.pdf).

Disclaimer: I'm one of the authors.

Re:Blockchain's Numbers Don't Add Up by nix0n on Thursday September 08, 2016 @12:55PM (#12454751)
Thank you, that's very informative.