Quote from: Come-from-Beyond on December 03, 2015, 01:06:44 PM

Quote from: Komputor on December 03, 2015, 12:55:41 PM

Question for CFB -

I've been trying to visualize the way the DAG would evolve in real world usage and how the transaction confirmation times would behave.
So if i understand correctly, the tips with the highest weights are encouraged to be approved first by incoming fresh transactions. In such a case, if there are is a sudden surge of new transactions followed immediately by a period of very low transactions, then many of the transactions in the first wave would end up waiting longer than usual for approval/confirmation, correct?

I then understand that the confirmation times may actually be very unpredictable in the initial stages of the network when there is a sporadic rate of new incoming transactions.
If true, then do you expect these uneven transaction confirmation times to even out as the volume of transactions eventually increases?

I'm no tech expert, but would like to understand this a little.

**Still reading the whitepaper**

It's a question to mthcl, I think. Simulation shows that Tangle pulses: its width, tx confirmation time and other parameters have clear cosine pattern.

Interesting, maybe mthcl can comment more on this. Thanks.

Also if you could clarify my second set of questions: So a strategy for devices stuck with unconfirmed transactions would be to send another blank transaction referencing the original unapproved transaction. Since this network will be used primarily by non-humans, such a behavior would need to be coded into the IOT device itself right? For example: If transaction does not confirm withing x time, then resend a blank transaction with reference to first one.
How would this work out in a larger simulation where many devices would use such a strategy? Could this cause a spam of blank transactions in the network where many devices are trying to promote their own transactions for faster confirmation times? Or is this a non-issue?