The author describes the database the nodes maintain, and the consensus algorithm in great detail (which is much much more complicated than the consensus algorithm of any blockchain in the industry), but does not say a word about scalability. You know: absolutely basic stuff, like are nodes full nodes? Is the history maintained in each node? How much storage a node needs? How fast a transaction is confirmed? No word about it in the whitepaper of a 10 billion project. I would not need it to be included if it were trivial. But it is not trivial. In interviews they claim that it is almost infinitely scalable. Followers spread the information that it is faster as there is more usage on the network. They say that millions of small embedded hardware nodes can participate and the network enables microtransactions, which requires incredible scalability. These are ridiculous claims, and it is incredible that there is nothing about it in the whitepaper, or any easily accessible document. They intentionally don't want to explain the technology in clear and structured fashion. Yes I did'nt spend a week of my time to try to understand the technology. Maybe a week would be enough to browse the source code and have answers to my questions. But I don't have that much time for IOTA. And yes, it is fishy, that no one tries to explain it in a way that experienced engineers could have a grasp of it.
Do you mind if a write a version of the comment you've just written (making it a bit less harsh) to the author of the white paper, and quote here any replies I receive?
(I also notice you list a contact in your HN profile, so I'm happy to include you on CC - but I think we should share any eventual response to your questions in this thread after we get it.)
Yes go for it. A well articulated communication on technical details of IOTA scalability is needed, and I think the whitepaper need to be upgraded with such information.
in case anyone is reading this thread later: per the above, I wrote the author a lightly edited version of nadam's question and a summary of this thread. He replied as below, offering to connect with people who know more. I asked nadam if he was interested but didn't receive a reply. So I didn't follow up except to say "thank you" and let the author now I've copied his comments into the thread.
Author's reply:
------
Hi,
thank you for reaching out to me, and for your interest in IOTA. You are right about the Tangle paper - I've put a lot of effort in it, and also tried not to overcomplicate it. But, well, it's "make things as simple as possible, but not simpler" thing - nothing comes for free, and so IOTA is really much more complicated than traditional cryptos, one simply cannot understand it without investing some effort first.
Regarding your questions: first of all, I'd like to stress that that paper is more about the Tangle - an idealized mathematical model, than about concrete details of IOTA's implementation. In view of this,
- are nodes full nodes?
In the paper it doesn't matter, there the nodes are just entities that sign the transactions and (more importantly) choose where these transactions will be attached.
- Is the history maintained in each node?
In principle, yes. Note, however, that the tip selection algorithm only uses the recent history of the system (that cumulative weight calculation only depends on this). This opens possibilities for periodic snapshotting: for the system to work, a node should know that recent history, and the account's balances before.
- How much storage does a node need?
- How fast is a transaction confirmed?
This depends too much on the concrete details of the implementation. Unfortunately, I'm not an expert there, I'm too much a "theory" guy. Should I connect you with someone more qualified?
