Welcome to the Science of Sport, where we bring you the second, third, and fourth level of analysis you will not find anywhere else.

Be it doping in sport, hot topics like Caster Semenya or Oscar Pistorius, or the dehydration myth, we try to translate the science behind sports and sports performance.

Consider a donation if you like what you see here!

Did you know?
We published The Runner's Body in May 2009. With an average 4.4/5 stars on Amazon.com, it has been receiving positive reviews from runners and non-runners alike.

Available for the Kindle and also in the traditional paper back. It will make a great gift for the runners you know, and helps support our work here on The Science of Sport.

Tuesday, August 31, 2010

The Sub-2 Hour Marathon? Who and when?

The two-hour marathon:  Who and when?  If at all...

This morning, I came across what has been a very topical point of debate for the last couple of years - will the two-hour marathon be run, when will it be run, and by whom?

The paper is featured in the Journal of Applied Physiology, and it asks these questions, and presents certain physiological concepts and questions that would need to be addressed in order for this performance to be possible.  Ultimately, the paper provides more questions than answers (because frankly, there are none), but it's interesting as a piece of a debate that comes around fairly regularly.  So today, I'll look at only one of the aspects used in the debate - previous performance.  The physiology of what a 2-hour marathon involves is for another time (otherwise this would be a mammoth post)

Back to 2008 - the barrier comes into view

Way back in October 2008, the debate began when Haile Gebrselassie ran 2:03:59.  As we are prone to do, when a barrier gets broken, we look ahead to the next one.  Gebrselassie himself was saying that he'd run 2:02 soon, which seemed a little optimistic given that he'd just taken 26 seconds off - so even another 2 performances of the same magnitude wouldn't have seen this happen.

In this case, we jumped four barriers ahead and asked whether it was possible for a human to run under 2-hours.  You can read my thoughts on this topic here.  David Bedford, race director of the London Marathon, was quoted as saying Without doubt I will see a two-hour marathon in my lifetime. It might be towards the end of my life. It might be another 20 years. But, yes, it will definitely happen."

Well, unless Bedford has figured out a way to live to at least 120, = it was suggested he would be wrong.

The paper - different approaches to the question

So looking a bit more closely at the paper, the first approach adopted was to analyse the progression of the marathon world record over the last fifty years and then to predict forward when it will dip below 2 hours.  This is of course a hazardous estimate - it was once predicted that women would outrun men in the marathon because women's times were dropping faster than men's times - past performance can be a very unreliable as a predictor of future performance!

Nevertheless, the paper makes the following observations, based on the graph below, which looks at world record times since about 1930:

So the drop in the 1950s and 1960s was pretty rapid (attributed to changes in training to a high volume, year round change, and African runners entry into the marathon event), followed by a much more consistent period of decline.  So on average, the record has fallen by 20 seconds per year since 1960, and so there are two scenarios:
  • If this rate continues, then the sub-2 hour marathon will be achieved by 2021
  • If the rate declines to say 10 seconds per year (based on the projection from 1980), then it will be achieved by 2035, 25 years from now
The third possibility, not mentioned in the paper, is that the decline drops right off, and settles at say one world record every five years, with each one taking 15 seconds off the time.  If this happens, then it takes 80 years to achieve, if at all.

You can make up your own mind over which of these projections are likely to be realistic.  It depends, of course, on whether you think the ceiling is being approached or not. 

One thing I will point out is that between 1988 and 1998, the record didn't move - it got "stuck" at 2:06:50 by Dinsamo of Ethiopia.  Then the unlikely source of Ronaldo da Costa broke it, and there was a spate of records.  So at the risk of stating the obvious, there may well be an average decline, but the real story is told by the "stagnations" and the "steps", and when asking whether a sub 2 hour is possible, you have to ask whether they will continue or not?  And then there is the obvious issue of whether 10 seconds per year is realistic?

For example, let's do the same exercise for the men's 10,000m world record (the reason for the choice of event should become clear later).

Below is a graph showing the drop in World records since 1950.

So, from 1950 to 2010, the average decline in the record is 2.8 seconds per year.  But, from 1977 (the first African WR over 10km, incidentally), it's only 2 seconds per year.

But then, very interestingly, it increases from 1993 again - you may decide what drove that (EPO, perhaps?  Better training?  More prize money?  Greater athletes like Tergat, Gebrselassie?) - to 2.4 seconds per year.  Then, since the current record was set in 2005, it's remained largely unchallenged.  And with the dearth of high quality 10,000m races on the circuit these days (outside of major championships), it would seem likely to remain that way.

In any event, apply the same method to the 10,000m record and let's ask whether we will ever see a sub-26 minute 10,000m?

If you take the overall average of 2.8 seconds improvement per year, then it predicts that we will see a 25:59 in 2011.  Ok, so that's not going to happen.

If we take the average progression since 1977, then the prediction is that we'll see a sub-26 time in 2013.  Again, I feel pretty confident in saying this is not going to happen.  The record has been improved by only six seconds in the last 12 years, and I can't see much happening from here on.  The point is that this kind of projection almost always under-estimates the likely progression.

Why 10,000m times are so vital:  Predicting UP

So now, why is the relative stagnation of 10,000m times important to this debate about the marathon? 

The reason is that the golden-era of 10,000m running in the 1990s, where the records fell so regularly from 1993 to 1998, is one of the main factors responsible for driving the drop in marathon performances in the 1990s and 2000s.  Regardless of the factors that drove that spate of records in the 1990s, it was the conversion of these runners to the longer distance that helped to lower the marathon world record and raise the standard to a point where a 2:05 now longer even guarantees a win.

The reality is that there is a strong correlation between the speed an athlete possesses over 10,000m and their ability to run a fast marathon.  It's not perfect, otherwise the best 10,000m runners would win marathons all the time (Zersenay Tadese, for example, has not converted his amazing 10,000m and half-marathon achievements yet).

And of course, there are exceptional marathon runners who have not had quite the same success over 10,000m (often due to circumstances and choice).  However, the basic premise is this - to run a sub-2 hour marathon, there is a cut-off limit to how SLOWLY you can afford to run over 10,000m.  And right now, there is no athlete with the 10,000m performance ability necessary to run at 2:50/km for a marathon.

The question is, if a 2:03:59 marathon is the result of a ± 26:30 to 26:45 10,000m runner (and this range is not that large), what kind of speed over 10km will be required for an athlete to run a 1:59:59?  And this is a vital question, for while I appreciate that there are many considerations that affect how shorter distance speed is translated UP to longer distances (the best 1500m runner is not the best 10,000m runner, of course), we do know that 10km and 21km performance do very closely predict marathon performance.  There are a number of reasons for this, mainly related to the relative speed sustained and the physiological/energetic processes for the two distances.  These issues are covered in the Joyner paper, which I will look at in due course.

So, if we apply this same logic to a half-marathon - the very best runners in the world are presently able to run a half marathon in 59 minutes.  They go through half-way of a marathon in 62 minutes, a 3 minute difference.  To run a 2 hour marathon requires 60 minute halves, and that athlete thus probably needs a half marathon best in the range of 57 minutes.  Is that achievable?  Certainly not now, but if you want to know when a sub-2 hour marathon is possible, the answer is "when an athlete can run a half marathon in 57 minutes".  And that is a lot further away than 13 or even 25 years.

A highly unlikely scenario of 10 seconds per year

My point here is that predicting marathon performance based on past performances may not be the best way to go - it's certainly incomplete, and what is missing is the recognition that shorter distance performance may be an even better predictor of marathon performance than previous marathon performance!

So, based purely on performance analysis, I would have to say that the projection that the record will continue to decline by 10 seconds per year is highly unlikely. 

Instead, I would be surprised if the world record is broken by more than 15 seconds once every five years.  Why?  Because that record is now so strong that the very best runner needs the very best conditions, pacemakers, race situation and course in order to challenge it.  So, 15 seconds every 5 years is, in my opinion, a best-case, and that gives us 80 years to wait for the 2-hour marathon.

And there is a problem in that the supply of great courses is now dropping below the demand.  For now, Berlin remains the prime hunting ground for records.  Dubai has established itself as super fast, and Rotterdam has re-emerged as a super fast course.  But I do believe that as that record gets closer and closer to the limit of what is physiologically possible, courses like London (surface, wind, competition and turns), Chicago (weather) and possibly even Berlin and Dubai eventually will require too many chips to be in place. 

Too strong a field (as in London) and the pace in the middle is likely to be erratic as guys watch each other and not the clock.  And remember how rain in Dubai scuppered a record attempt?  Or a wind in Berlin?  It's now too precise, the margins for error too small.  So the notion of every race being a record-attempt will cease, and we'll see viable attempts (where "viable" is a guy on 2:04 pace at 35km) only once every 5 or 6 years. 

Of course, there is physiology too, and a large part of the Joyner paper is devoted to talk of running economy required for a sub-2 hour marathon.  This is maybe the most interesting part of the discussion, much like it was for the cycling analysis we did recently.  We'll dig out the analysis done on Zersenay Tadese, one of the most interesting things we've looked at on this site. But this post is already long enough looking only at historical performances, so I'll leave economy alone for now and return to it later this week!


P.S.  A copy of the PDF of the Joyner paper is available on LetsRun.