[reader50]

The Phlump vs Team Competition has had it's stats figures jump around a bit. Such things need to be explained in detail, or people may feel that there is cheating going on.

I've had to adjust and recompile the stats multiple times to cover irregular stats updates from dnetc during our race. This is unfortunate, because stats should be solid, and adjustments late in a race look very bad.

The net result has not affected Phlump's totals, but have benefited the Team totals by about 375K. Here are the details:

• This is a crunching contest, so only units crunched while people are on the team count. This is why the Team total listed on the Phlump competition table does not agree with the total listed in our Quick Stats bar for RC5. To maintain that, we have to keep an internal tally of units crunched since Nov 1, 2001.
• +115K After robertk merged his accounts on 12-8-2001, we apparently crunched over 2.5M for the day. I had been using the daily crunch totals of our members to avoid the irregular daily figures from RC5. With such a huge error in our internal figure, the competition stats had to be moved to the official Yesterday figures as posted by RC5. The difference is that new members' first day of crunching with us is counted in the official figure, but is not counted in our internal figures until they appear on our team for two successive days.
• +260K (estimate) when RC5 did not do the stats run on the night of 12-12-2001. This produced zeros in many of our stats pages. Later checking revealed that totals for the Team and Phlump caught up the next day, but the "crunched yesterday" figure did not include the missing day. I had to interpolate the missing figure. At the time we had been crunching about 290K per day, the next day showed 320K crunched, while later days dropped back again. I estimated that 30K of the missing work units got bumped into the next day's crunch figure, so my estimate was 30K lower to compensate. Ironically, our figures would have been ok if we could have remained on our internal member crunch totals.

I sincerely hope that RC5 will not throw us any more curve balls before the contest ends. For the moment, the Competition figures are accurate to the best of my knowledge.

- reader50
Team MacNN

[reader50]

Here we go again...

RC5 missed another stats run, on the night of 12-16-2001. They were down for stats maintenance for around 12 hours. I've put today's (yesterday's, now) figures on the site, but the data is suspect and will doubtless have to be rechecked over the next couple of days, as dnetc's stats settle back down.

[slipjack]

Thanks for the diligence... but we need more crunchers!

[reader50]

The Competition figures have been corrected again, since RC5 stats came back today. The Competitions page has been updated.

Change Log:

• RC5 stats were down for three days, there was no data on Dec19, Dec 20, or Dec 21. Note that their stats pages were up some of the time, but the data was frozen at the Dec 18 values.
• Team Active production across the gap was 897,288 units, of which 280,270 showed in our official Yesterday total. The difference (617,018) was divided by 3 and applied to each of the 3 missing days. Dec 19: 205,673, Dec 20: 205,673, and Dec 21: 205,672.
• Phlump's production across the gap was 703,485 units, of which 232,778 showed in his official Yesterday total. The difference (470,707) was divided by 3 and applied to each of the 3 missing days. Dec 19: 156,903, Dec 20: 156,902, and Dec 21: 156,902.
• The current gap is 733,112 units in Phlump's favor, with 10 days left in the race.

Note that cruched units on Dec 31 will count. I will use the data posted by RC5 on Jan 1 to determine the winner.

If reliable figures are not available on Jan 1, I will have to begin with the next figures that RC5 supplies, however many days later, and interpolate both parties' positions as of Jan 1. Should this be needed, it would be helpful if everyone continued crunching steadily on RC5 until then, to simplify the estimation process.

[slipjack]

I'm back to full power... everyone at work is off fer X-mas and were happy to let me install RC5 on their machines + my dualies aren't freezing using the older version, which helps.

[reader50]

The Competition table has been updated again.

Change Log

• On Dec 24, rc5 stats finished catching up for the missing 3 days (Dec 19-21). This did not show in the official Yesterday totals, but did turn up in member totals.
• Team production correction was an additional +146,053 units. This was divided by 3 and added to each of the dead days. (48,684, 48,684, 48,685)
• Phlump production correction was an additional +178,809 units. This was divided by 3 and added to each of the dead days. (59,603, 59,603, 59,603)
• As of this time, the gap is 795,246 units in Phlump's favor. Counting today, there are 6 crunching days left in the race.

[cdhostage]

Looks like we lost to Phlump. Will lose anyway.

Oh well.

I'm going on Ubero tonight.

[reader50]

Phlump is the winner. Congratulations.

The results have been posted on the Competitions page.

Dnetc stats have caught up to the present, and have been stable for several days now. Considering the large gap at the end of the race, I did not bother doing trend analysis in calculating the final figures. If I had, the gap would probably have been another 70K or so in Phlump's favor.

Cheers, everyone. It was a good race. Close ones usually are, and we came within 7% of winning.