Thread by @SarahJamieLewis

IOTA have released the whitepaper for their highly anticipated "Coordicide" aka "make IOTA not centralized". Have they actually solved the problem of a secure, decentralized, scalable fast distributed ledger?

No, of course not. They've added yet another layer of indirection.

They have solved the problem relating to peers not always having a global view of the network by assigning them a value which is derived from a function requiring a global view of the network

They have totally solved that problem too don't you know. Eclipse attacks are impossible if decisions are local and "unpredictable"

"Ah, but Sarah", I hear your exclaim, "didn't IOTA also propose an additional layer of security!?"

By which I assume you are referring to their probabilistic voting proposal, where votes are weighted on the Sybil "protection" mechanism above.

This is just amazing.

LOL did not take long for the IOTA bandwagon to roll in. So let me just say 1) yes I read the whitepaper 2) yes I read the AMA

This is the same story as every other IOTA project, going back years now. The problems are never solved, they are simply renamed & abstracted.

Remember that time IOTA just flat out took possession of tokens that were deemed "at-risk" because of a vulnerability, and then required an ID check to get them back?

https://twitter.com/SarahJamieLewis/status/1012918320931090432

I have my own theories on what is actually going on under the hood at IOTA, based on their leaked meeting transcripts, board member resignations and the kinds of partnerships they are forming.

I hope someone does a full investigation one day.

Anyway, I'm just going to restate the current truth as we best understand it. If you want to decentralized, scalable and fast you will definitely almost always compromise on the decentralized part (either through process or practice, through policy or payment).

There are real, legitimate conversations to have about decentralization means, and what compromises are acceptable (e.g. see @angela_walch 's Deconstructing Decentralization papers.ssrn.com/sol3/papers.cf…)

IOTA isn't contributing to those conversations, they are selling snakeoil.

Oh, for the sake of disclosure. I own and or control by proxy some more than 2 of: Bitcoin, Monero, Zcash and probably still some (m)IOTA.

The IOTA I have was obtained purely accidentally, because the 'S' in IOTA stands for security.

"control" in the same way that I "control" all the email accounts whose password is "password"

Just gonna leave this here too

https://twitter.com/SarahJamieLewis/status/999905051073069056

Oh I also own more than 1 ETH and REP, I keep forgetting about them.

"you didn't say anything"

There is literally nothing to say, this isn't highly technical cryptographic algorithm with a subtle flaw, this is a solution built on the foundation of a sybil protection mechanism that *cannot* work because it itself relies on true global state.

One of the goals of IOTA is a highly distributed network which is *forbidden* from having global state *by design* (see CfBs' writeups on IOT bandwidth and 5G).

You can't solve "nodes can conceal transactions from other nodes" through transaction-based reputation

All you do is create a meta-problem "nodes can conceal reputation transactions from other nodes"

The only valid assumption you can introduce to solve that is that "transactions eventually get delivered to all nodes in a reasonable time".

(p.s. now you have a liveness problem)

Of course you can solve that by strictly defining the number of nodes that are needed to participate in such a protocol (fewer nodes == more believable and actionable assumption). Oh but wait, tread carefully because now you are defining which nodes have canonical state.

I'm coming around to the belief that this place is perfectly fine position to be in (it's effectively analogous to the inherent kind of centralization that come with mining pools), but projects need to be very upfront about that very real centralized dependency.