If I run a task for 36 hours, and somebody else runs it for eight, we'll have completed a vastly different number of models.

But our results can be used to verify each other's contributions.

What's the context behind why this works?

But our results can be used to verify each other's contributions.

They can't, and they're not.

A Task is sent out once. If it Validates, that's it.
If not, then it is sent out a second time and that result is run through the Validator. If it's good, then that is the result that goes into the science database. If not, then the Task is determined to have too many errors & that's it.

---

Grant
Darwin NT

I'm running some _1s that have many more models completed than the original work.
When they finish, they're validating against contributions with fewer models.

You can track how many tasks your partner has run by clicking on the 'work unit' button on your results page, then the 'task' that isn't yours.

For example, this person contributed 209 decoys.
https://boinc.bakerlab.org/rosetta/result.php?resultid=1593187432

Its pair, on my computer, has reached model 4300 and counting.
It has another 14 hours, 40 minutes to go.
Here's where you'll see the results: https://boinc.bakerlab.org/rosetta/result.php?resultid=1593314585

I'm running some _1s that have many more models completed than the original work.
When they finish, they're validating against contributions with fewer models.

No they are not.
If that were the case, then every single Task would be sent out twice straight away- and that doesn't happen.

If a Task is sent out a second time- if the late Task is returned before the resent one is started, the resent one is Cancelled by the server.
If it has already started processing, then when it is returned it is Validated- with no comparison in any way, shape, or form with the previously issued version of that Task.

---

Grant
Darwin NT