WCG has set HFCC to normal weight in the total distribution and DDDT to low weight, so I'm getting about 1 in 8, 1 in 10 when choosing this mix. This has not been changed since the restart, which produces about 13-14 CPU years per day for this project or 20,000 results.

You can find out how much work is being done on the various projects by clicking on the project links on your "My Grid" home page. (You can also get to these via the "Statistics" button at the top of all WCG pages, and then "By Project").
Look at the numbers under the "Yesterday" subheadings.
Note that these numbers are based on results returned, so they lag some hours or days behind work sent out, and they average out short-term variations in the rate of work sent out.
Example: For Yesterday, 6 Apr 09, CPU hours -
DDDT: 14 yrs, 316 days; HFCC: 33 yrs, 89 days; FAAH: 55 yrs, 207 days

I think you were just unlucky in getting so few DDDT, but keep in mind that computers that are set to get work from just one project get first pick of the work for that project. Thus when work for a project is in short supply like it is for DDDT right now, there may be very little work left over for computers that have more than one project selected.

Growls don't get us anywhere without some background info. Did your client ask for work and is it DDDT only and did you get a

15/06/2009 10:02:50 AM|World Community Grid|Message from server: No work is available for Discovering Dengue Drugs - Together

For,

I changed my profile to only get DDDT, upped the cache setting by 0.2 and immediately had

15/06/2009 17:50:17 World Community Grid Finished download of dddt.15.dat.gzb

10 day stats, including projection for today:

Jun.06-09 14499
Jun.07-09 14445
Jun.08-09 14301
Jun.09-09 14630
Jun.10-09 15696
Jun.11-09 17990
Jun.12-09 19111
Jun.13-09 19851
Jun.14-09 19538
Jun.15-09 14989

Maybe back to the normal pace, for this project, but their FLU is still up, also from Galveston.

May have been but in hind-cast, the curves now indicate there was an undisclosed ** supply issue of sorts. The signs are spike in run time and drop in credit per hour, followed by a dip back of the run time and spike in credit per hour. Sort of slow machines catching up, then fast machines running away again. In these cases, those on preferred project always get first served on the remaining and new work.

** Maybe better to say, unexpected. The techs have more things to do than report on every little vibration in the grid. Think there are by now with 4 platforms and different homogeneous redundancy groups, like 50 streams of results to maintain and supply.