Support for once-a-week connections

Some people have machines that connect just once a week to the net. There are two issues for such machines on the current set of policies.

In suggesting ways to address these issues I am expressing wishes, not intending to be critical of the project policies.

1) With WU timing out after 7 days, small random errors in the timing of the weekly connections mean that maybe 50% of the WU are deemed late by the time they are returned, costing the project resources if replacements have already been assigined

2) Issues of how many WU can be crunched in a week - many more than the 8 that the project currently allows.

My suggestions to address these are

1) an 8-day life for WU. Would only cost a little extra in terms of holding WU for longer, while adding to the potential donor pool all those with a 1-week pattern of connection.

2) At present the system allows 8 WU as the startup value, but deducts 1 from your allowance each time there is a non-return or an error, down to a minimum allowance of 1; and doubles the allwance whenever you return a valid WU, up to a max allowance of 8.

My suggestion would be to allow the increase to continue (on a plus 1 basis, not doubling) from 8 up to some higher limit.

Even better would be to make the limit depend on how many WU returned in the last week (say last week + 8 WU).

In this way permanent rogues would not get more than 8, while a high perforoming machine would be allowed to process the workload it had shown itself to be able to handle. The damage done by a good machine that turned rogue would be limited to something around the scale of the good that that machine had already done.