Decentralization is not binary
When building a decentralized protocol, there are three primary design choices for the creators to consider:
How does power get distributed within the network?
How is consensus achieved in the network?
How do changes to the protocol get implemented?
Satoshi's approach to decentralization
Bitcoin is the most decentralized protocol that's been created to date. Power (via tokens) was distributed by Satoshi publishing a white paper, releasing software and allowing anyone to mine at the onset with complete transparency of the rules. Consensus is achieved by anyone running the software and expending electricity. And protocol changes are deliberately made and require consensus from the core devs. This approach maximizes security (there's no central points of failure or major security bugs) at the expense of speed of iteration and innovation.
Vitalik and @dantheman's approach to decentralization
Some in the Bitcoin world argue that anything less decentralized than Bitcoin is not decentralized and as a result doomed for failure. They're missing the fact that decentralization is not binary. While protocols like Ethereum and Steem are not as decentralized as Bitcoin, they are more decentralized than the existing systems they are disrupting and they have the potential to grow more decentralized as they mature.
Decisions about decentralization require one to evaluate the tradeoff between security robustness and speed of iteration. I believe there is a strong reason for both ends of the spectrum to exist: fast interation at the expense of security (Ethereum, Steem) and aversion to change with the benefits of security (Bitcoin).
I'm not sure Ethereum or Steem got the tradeoff right; it's possible that they sway too far on the side of centralization and something will come after that gets it right. But I'm glad they're trying.
The decentralized protocols that get these tradeoffs right will be successful.
Trent from BigchainDB wrote a really great article called The DCS Triangle which suggests decentralization, Consistency, and Scale being the three things different protocols can optimize for. As you say, there are also degrees of each of those, and the article gave me a good understanding of which trade offs are even possible.
Trent also reviews #IPFS in his write up. I'm very interested in what happens with it as it doesn't have anything related to currency built into the system.
Good stuff thanks for sharing.
As far as I know, IPFS is struggling on performance because there's not enough nodes running in the system due to lack of incentives. So they're launch a token (filecoin) that they hope will change that .
There will be a variety of incentives around the base protocol. I like that's not tightly coupled to currency. It makes IPFS a natural complement to a variety of other systems.
You can't really call a protocol "struggling" -- adoption and resiliency are issues, for sure, but the inherent nature of the protocol make it very well suited to sharing certain types of information.
For instance, posting images and other binary files to steemit via IPFS could be very interesting. The content hash would be encoded with the blog post, and could be served a number of different ways.
Why can't you call a protocol struggling? IPFS is struggling on performance. Services actively aren't using the IPFS protocol because of lack of speed and uptime. Again, once filecoin is launched the hope is that will change because more IPFS nodes will have incentives to run in the network.
Would you say HTTP is struggling? Someone might not be running enough Apache servers, but the protocol is just fine.
I get your point for decentralized systems, it's actually a mix of the protocol AND active nodes. I'm excited by IPFS-the-protocol, because it's a non blockchain based system, and we're going to need a mix of different approaches.
Sure, you'd say HTTP was struggling if it wasn't getting adoption.
He, Bitocin is not decentralized. Bitcoin is way more centralized than ethereum or steem. Since few Chinese miners are owner of bitocin, how can you say bitcoin is decentralized?
The hash rate distribution looks pretty distributed from my perspective right now. No miner has more than 18% of hashrate (see below, source here)
If you haven't read Dan's previous article about Bitcoins security and the mining pools, I kindly suggest you give this article 10 minutes of your time to read through it.
http://bytemaster.github.io/update/2015/09/29/Bitcoin-is-100x-less-secure-than-commonly-believed/
While reading the article, keep in mind that the mining pools are a central point in the network.
There's no central point of failure in bitcoin.
There's a pretty clear central point of failure in Steem currently.
There is central point of failure in bitcoin.
Centralized mining. steem is 100x times secure than bitcoin in terms of central point of failure.
Wow, look at that chart 17.9% f2pool!
That chinese miner guy can collapse bitcoin anytime.
If you are using bitocin, you are under the generosity of super elite of CCP.
That f2pool {w,sh} ould do whatever super elite of CCP says.
Satoshi knew if mining centralized, bitocin would die. So he fought to last for survival of bitcoin. He even suggested gentlemen's agreement to distribute hashing power at desperation. After serious centralization he knew it is failed, so just disappeared.