I see that on some models the initial replication is sent to 5 or even 10 hosts. How many of these hosts are needed to complete the model for the science be counted? Just 2?

I am asking because on some WU I see that at least 2 other computers are successfully returning trickles and does this mean that in this case my computer running the model is redundant?

The number required depends on the researchers.
In some cases they set a high number because they\'re really interested in the result for that set of parameters. (Perhaps because it was sent previously, and a partial result contained something interesting.)
In other cases, it\'s because they\'re conducting an integrity check on the data being received.
It may also be because they want to see what the difference is between models when the same dataset is run on different types of computers.

Please continue running any dataset you receive until the model has run as far as it can.

---

Backups: Here