Haile Gebrselassie 2:06:09 in Dubai

Fast for Geb in Dubai, but the record was never on

2:06:09 can never be deemed a disappointing performance - it's a time that ten years ago would have scared the world record, been comfortably in the top 10 of all time, and would signal a great effort.  It's testament to how the marathon has progressed, and to what Haile Gebrselassie has achieved in his running career, that when he achieves this time, it's met with some disappointment.

Gebrselassie spends most of his racing time hunting world records - I've said before what a pity this is and how great it would be to see him take on the likes of Lel, Wanjiru and Kebede, but he has been content, for the last five or so years, to race the clock in a couple of races a year.  His first attempt for the last few years has come in Dubai, where huge incentives, both upfront and for winning and breaking records, are a big draw.

And Dubai has been very fast for Gebrselassie - he has run 2:04:53 and 2:05:29 (2nd and 8th fastest ever at the time, respectively).  But 2010 has produced the slowest of his three Dubai Marathons, 2:06:09.

Dubai 2010 - close for a short time, then a gradual slide away

And the record was on for a short time only.  The graph below shows the kilometer splits, and then I've worked out the projected time at 5km intervals (based on the cumulative time, not the last 5km, note), shown in green where he sped up over the interval, red where the pace slowed.

Remember that the world record of 2:03:59 requires an average kilometer pace of 2:56.  Gebrselassie was close to it, but never below it, until 5km had been covered, and his 5km time of 14:54 was 3 seconds per kilometer off that required pace.  We've discussed before how incredibly precise the pacing has to be in order to succeed over 42.2 km - it is remarkable.  In previous Dubai races, Gebrselassie has gone out much too hard - in 2008, he was on course for about 2:02 after 15km, and then slowed considerably to run 2:04:53.

This time, if there was a pacing error, it was on the slower side of desired, because the early pace was just outside the target - Gebrselassie averaged 2:59 for the first 5km (14:54 split), 2:58 for the second 5km (10 km split of 29:42), but then began to slow progressively.  At halfway (1:02:51), his projected time was 2:05:42, which some would say meant the record was still on, but when you need to run the second half of a 2:04 marathon in 61:07, then the record is pretty much gone. That the pace was slowing after 10km is a sign that it was never going to happen today, and from that point onwards, only eternal optimism would have kept record hopes alive.

A race in the final kilometers?

The pace held steady at around 2:58/km from halfway to 35km, and then the pace really did slow down - a series of 3:05 or slower kilometer times and the projected time dropped outside 2:06.  It is very interesting to note that Gebrselassie's final kilometer was 2:36, an increase in pace which suggests that he took the foot off the gas in those earlier slow kilometers, because by then the record was clearly gone - it would have required something in the range of 2:50/km for the final 10km to challenge his mark.

Also of interest though, is that he "only" won by 24 seconds today, making this the closest marathon "race" he's run in a while.  He may have entered looking for a paced record, but he seems to have found himself in a tight battle at the end.

And, given a final kilometer of 2:36, the 24 second winning margin would have been largely, if not entirely, created in the final kilometer, which means that Gebrselassie was not alone at 40km, and definitely not at 30km.  It seems likely, then, that the presence of other runners in a group from 30km onwards, combined with Gebrselassie not feeling quite up to the world record (as shown by the relatively sluggish start, particularly from 10 to 20km), resulted in a tactical race which slowed the pace down in the latter kilometers.

I confess that I didn't see the race, it wasn't televised and SA internet is so poor that watching online doesn't happen at the best of times, so I don't know the circumstances in those middle kilometers.  The temperatures and wind certainly don't seem to have been too limiting - the temperature at the finish of the race was 18 degrees with no wind, though the humidity was up at 80%, which may have contributed to that progressive decline.

Nevertheless, it's a good, not great time, for an athlete who has inspired expectations of great every time he runs.  It really doesn't help that he only runs in time-trials (with the exception of Berlin last year, where Duncan Kibet might have, but didn't, provide some opposition), and that he talks up the chances every time he races.  Before this one, he spoke of perfect preparation, and the ability to run 2:03:30.  There's always the proviso that the conditions must be absolutely perfect, and so I suspect in the aftermath of this race, there'll be some problem with the conditions, perhaps that humidity.

We'll see, and update the post a little later, and maybe debate in the comments below!

Ross