Spoke to soon in according with Murphy's rule. The last 6 are of batch 301 and 302... 10 hours.

You've jinxed me now... There I was happily munching through batch 253 and since then I've got nothing but batches 301-303 now... all 10 hours. I guess from this that there may still be quite a supply of 10 hour batches until we start to see the 6 hour ones coming through in the majority?

Edit: Just had a look at Sekerob's graph from 2 days ago, which shows average runtime of about 9.65 hours per WU. So it looks like it might have only been a ratio of 1:11 for 6 hour:10 hour WUs then, but I guess we need to take into account varying buffer sizes to know how recently those were sent out, but the 10 hour ones are still well in the majority until that number drops below 8

They will come out mixed for the next week or so. We should finally settle into the 6 hours workunits next week.

Finally got a 6-hr WU, immediately followed by 2 more 10-hr jobs. I still think my machine is a magnet for the longer units (LOL)

According to Sekerob's Rice Graph everybody is still getting 10-hour WUs these last days!
The problem with getting rid of these longer WUs is that Rice is our recommendation for slower machines because of its fix duration which protects them from getting very long WUs.
This is all fine and I would not change my mind on this recommendation, but as a consequence each WU returned from a slow device is not making a huge progress towards the goal of computing the desired number of seeds for each batch. Hence more 10-hour WUs have to be generated and distributed to reach the target.
But this target is fixed and we will complete these 10-hour batches sometime anyway, we just have to be a little patient.

Cheers. Jean.