5

https://blog.iota.org/a-primer-on-iota-with-presentation-e0a6eb2cc621

On the presentation Dominik describes the transaction initiation process, in which the second step is random tip selection, where a tip according to the definition is an unconfirmed transaction. On the diagram the tips are colored in grey, and they don't have incoming arrows. If every participant in the network uses the same MCMC algorithm to choose two random tips out of all sets of tips, then no node should have two incoming arrows.

enter image description here

On the diagram all green and almost all red of nodes have multiple incoming arrows, which means the MCMC choose nodes that have already been confirmed by at least one node. If I am reading the picture correctly, the tip selection algorithms does not select only tips, but transactions that have not been fully confirmed. That is a new node runs a MCMC algorithm to choose two previous transactions to confirm out of the set of grey and red transactions.

0

1 Answer 1

3

Not all nodes see the same state of the tangle at the same time. So it can be expected that two nodes pick the same tip (which might actually even not be a tip anymore) at the same time and issue their transactions building on this node concurrently.

It is not a problem if some transactions are directly confirmed by two others. However, on average a new transaction should confirm at least one transaction not confirmed by another transaction, otherwise the number of tips would increase indefinitely.

1
  • Understood. So each node (well behaved node) runs locally a tip selection algorithm, but because the tangle is distributed and the confirmation takes time (PoW) nodes may choose the same tip and we end up with multiple direct confirmations for the same transactions. I find this peace is not explained well, as on all diagrams and animations a new transaction is selecting two true tips and is confirming them, when in practice that is just a local view of the tangle.
    – serega
    Commented Dec 8, 2017 at 17:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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