Some people argue that Iota is centralised because of the use of a coordinator to issue milestones which, in turn, are used as a root of trust to verify transactions.

What would happen to the network (with the current IRI) if the coordinator vanishes?

  • 1
    The Milestone is initialized to null I think, and until a Coordinator issued transaction is found, all TransactionsToApprove returns a null list. So I think the network just grinds to a halt. Of course, there is one sure way to find out ... just spin up an iri and interact with it on a private network. – Austin Powers Jul 22 at 10:57
  • Keep in mind that in the scenario suggested by the OP, old milestones still exist in tangle. The question is "what append if there is no new milestone issued ?" – ben75 Jul 22 at 14:17
  • Ah right. The synced tangle from a full node would provide the old ones. – Austin Powers Jul 22 at 21:04
  • So it would just slow things down by not caching them outside of the database. – Austin Powers Jul 22 at 21:21
up vote 2 down vote accepted

No more confirmation by the coo, the tangle will keep growing, but none of the new transactions will be confirmed.

With the current IRI (1.5.1), the starting point of the random walk is a "not too old" milestone. If there is no more new milestones the random walk will probably became more and more expensive to run. So the selection of transactions to approve (on regular IRI) will slow down. I guess that most instances will start to crash by lack of resources (I expect that the RAM required by the random walk will be the bottleneck).

After reboot, the IRIs will rebuild a new tangle using the latest milestone as a starting point... and crash again...

There won't be any 100% confirmed transactions anymore. The tangle will then rely on the fact that the majority of people will follow same path in tangle as COO referenced before.

Random walk should not become much more expensive here as well, since the time when IOTA is ready to remove the COO it will be such a high number of transactions ( and thus network bound pow will be very high ), that an attack (side tangle , double spend) won't be a trivial task.

see also

The value alpha for example helps to calibrate confirmations rate in that scenario. A high value of alpha would help to strengthen the confirmation rate and reduce the amount of time spent in random walk.

  • It might also be that with such a high number of transactions, that an attack (side tangle , double spend) will be a trivial task. It can go both ways. Also ... the number of transactions required would be guaranteed to DDOS the network if nothing else (see What are the prerequisites for shutting down the IOTA coordinator?) I think it requires at least 26,000,000 TPS (minimum). – Austin Powers Jul 22 at 11:05
  • 1
    And what is "network bound pow" anyways? – Austin Powers Jul 22 at 11:05
  • network pound pow is the combined pow work the network is able to do. Since it is not possible to attack the network solely in one spot you would have to attack the network globally (with 34% of hashing power) in order to have a chance to build a new tangle. That will be increasingly difficult with a higher transaction rate of the network. – instantlink Jul 22 at 11:06
  • any insights on where the 26,000,000 magic number comes from? its not even trinary :) – instantlink Jul 22 at 11:18
  • 2
    There is no better theoretical number however. It may be wise for someone to answer the question with better numbers. – Cybergibbons Jul 22 at 19:27

Your Answer


By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.