How to break the marathon world record in Atlantic City, New Jersey

Does this sound like an impossible dream? Well, read on, and let me convince you.

We are currently living in a golden age of marathon running.  Paul Tergat's 2003 world record has been bested forty-one times in barely twelve years—an unprecedented occurrence, especially considering that top marathoners typically only race one or two marathons per year.  Performances that, only a decade ago, would be earth-shattering are now relegated to also-ran status.

There are a number of reasons why elite marathon times are routinely three to four minutes faster than a decade or two ago: more young, talented runners are moving to the marathon in their prime running years; prize money and appearance fees have never been higher at big-city marathons; track 10,000m events are rare and not profitable to run anymore; and top African runners and their coaches have discovered new training methods that push the body to find new sources of energy, creating a breed of what coach Renato Canova calls "turbo-diesel" runners. 

The relative importance of each of these explanations for the current banner crop of sub-2:05 marathoners is up for debate, but analyzing these is not my goal today.  Instead, I want to outline how current champions can go even faster.

It doesn't involve drugs, and it doesn't involve any changes in training.  Instead, making a number of alterations to the actual marathon event itself should do the trick.  Modern marathons are still conducted according to the logistical needs of a big-city event that caters to the general population, and sometimes, this doesn't result in ideal circumstances for running as fast as possible. 

Attempts at distance world records on the track are planned months in advance: the entire field consists of world-class runners, expert pacers are enlisted to ensure the pace is spot-on, and the venue is selected only after careful consideration of things like temperature, wind, and quality of the running surface.  In the future, marathon world record attempts will look very similar.

Big-city marathons that have played host to world record attempts, like Berlin, London, and Rotterdam, still have a number of deficiencies which could be costing elite marathoners precious seconds over the course of the race.

First among these is start time of the race.  Virtually all marathons start early in the morning; this is for weather-related reasons (temperatures and wind are lower in the morning), but also for convenience and safety for general participants.  A marathon that started at six in the evening would still have back-of-the-pack finishers coming in at midnight, so a morning start is a no-brainer for a mass participation event.  From a peak performance standpoint, though, this is suboptimal. 

The 6:30 a.m. start of the Dubai Marathon

A huge raft of research shows that performance in endurance exercise peaks in the evening, not the morning.  For example, a 1983 study by Baxter and Reilly demonstrated that performance in 100m and 400m freestyle swimming shows a consistent, inverse linear relationship between time of day and swim performance.¹  Performance at 400m (about equivalent to a 1600m run, duration-wise) was 2.5% better at 10 p.m. versus 6:30 a.m.  As the authors concluded, early morning was clearly the worst time to schedule an all-out endurance test!

A lot of coaches and athletes have figured this out already—the fastest heats of the 5k and 10k at meets like Mount SAC Relays and Stanford Invite, where numerous NCAA and American records have been set, are always late at night.  And most high school and college runners have figured out that tough interval sessions seem harder when they're done at Saturday morning practice instead of at a typical weekday afternoon practice.

There are a few counterpoints to consider.  Some emerging research suggests that genetics plays a role in whether you are a "morning person" or an "evening person," which in turn affects your performance in morning versus evening endurance events.  It's unclear whether this is merely a relative effect (i.e. their morning performace does not suffer as much), or whether some people really are at their best in the early morning.  Still, a 2015 review of the current state of the science by researchers at the University of Bergen in Norway concluded that athletic performance is best in the evening.²

Some top Kenyan and Ethiopian runners might have morning-type genotypes, and thus would either be unaffected or negatively affected by evening marathon races.  Top runners might also not be as affected by time of day as the teenage swimmers in Baxter and Reilly's study.  Nevertheless, some top runners are surely evening types, and top talent can't circumvent all barriers.  Altitude, for example, allegedly affects Kenyans and Ethiopians to a lesser extent than Western runners, but they still can't run their absolute best unless they are at sea level.  

The ideal elite marathon would start somewhere between 6 p.m. and 10 p.m.  This would let elite runners take advantage of the circadian variation in performance, instead of fighting it by racing hard in the morning.  Obviously, training would need to be changed, too—tough marathon-specific workouts would need to be performed in the evening during the buildup to the race.

While we're considering the effects of the circadian rhythm on running performance, it's worth pointing out that Kenya and Ethiopia are eight hours ahead of Eastern Time.  For a marathon race in the United States, traveling such a great distance will induce serious jet-lag which can affect performance for several days.  According to Thun et al., westward travel induces deleterious effects on athletic performance for at least half a day per timezone, though there's considerable individual variability.²  If our world record attempt was to occur in the United States, this suggests that athletes from Europe and East Africa should arrive at least a week beforehand to be totally recovered from travel and jet lag before the race. 

The two enemies to running fast on the roads are hills and turns.  All world-record caliber courses today are very flat, but a number of them still have a lot of turns.  This is a necessity in old cities with haphazardly-planned streets dating back hundreds of years, and multi-loop courses are out of the question for general participation events—elites would lap too many slow runners.  Dubai is the only major marathon that's planned a course that largely avoids turns; their race is essentially an out-and-back with two hairpin turns. 

Even the blazing-fast Berlin course has quite a lot of turns

An easy way to best this is to use a criterium-style loop course.  Holding a marathon over a 4x10 km course makes it vastly easier to choose a perfectly flat route without excessive turns.  However, this necessitates that the event be an elite-only competition.  The United States Olympic Team Trials in the marathon followed a similar design in 2008, with the field covering four 10 km loops.  A loop of this length also has the advantage of being just long enough so that the fastest men (~2:02 marathon pace) would not catch up to the slowest elite women (~2:30 marathon pace), assuming the women started a couple of minutes after the men. 

Even better, there's no need to totally exclude the general running public from the event.  You could host a 10k road race on the marathon criterium loop, starting at 6 p.m., for example, then have a post-race party coinciding with the elite race (a full marathon) on the same course, starting at 8 p.m. 

A criterium loop also makes hosting an evening event a bit easier.  Instead of having to illuminate an entire 26-mile route with floodlights, you'd only have to cover a 10 km loop. 

Most marathons are held in temperatures that are too hot for elite runners.  A 2012 study by Nour El Helou and other researchers in France examined the relationship between race-day temperature and performances at six major marathons over the course of ten years.³  Through statistical analysis, El Helou et al. demonstrated that faster marathon performances require colder ambient temperatures.  This makes sense, as running faster generates substantially more heat.  So, the optimal temperature for a 2:02 or 2:01 marathon is colder than what's comfortable for a three or four-hour marathon. 

The differences in optimal temperature are not negligible, either—back-of-the-pack runners tend to run their fastest at about 47° F, whereas El Helou et al.'s data suggests that a 2:02:00 marathon calls for temperatures of 34° F!

Outlier data from top 1% women not included.

For an elite women's time, it's a bit harder to predict optimal temperature.  El Helou's data showed a significant aberration in the optimal temperatures for the top 1% of female marathon finishers.  The data suggest that top women require unusually warm temperatures compared to slower female marathoners, and even compared to men who run the same speed.  It's uncertain whether this is just a fluke in the data or whether it's a real phenomenon.  I contacted the authors of the paper, but received no response.  One possible explanation is that elite female marathoners are very small, and thus lose too much heat in cold temperatures, on account of their very high surface area to mass ratio.  I'm planning on trying to replicate El Helou's results using data from other marathons, and will try to update this section once I've done so.

Another important point to emphasize is that the linear trend may not hold beyond a certain point.  El Helou's data only goes down to about 2:40 marathon pace (which calls for an optimal temperature of 39° F); it's uncertain if Kenyan and Ethiopian marathoners would benefit from colder temperatures.  Again, the surface area to mass ratio issue could prove to be a problem.  Happily, this sort of question can be answered relatively easily in a physiology laboratory, assuming you could recruit some top athletes to serve as volunteer subjects.  Regardless, it's unlikely the optimal temperature for pro male marathoners will be above forty degrees Fahrenheit.

Too often, distance record attempts are foiled early on by suboptimal pacing.  In marathons, usually this means races getting out too fast.  The 2009 London Marathon, for example, got out in 14:08 for the first 5k.  Seeing as 2:03:00 marathon pace is 14:34 5k pace, any attempt at a world record was ruined just fifteen minutes into the race.  As noted by some clever analysis at LetsRun.com, distance world records essentially mandate running dead-even or slightly negative splits—deviating from the goal pace by no more than three seconds per kilometer throughout the entire race.

There are two easy solutions to make sure the rabbits (and there will need to be plenty of them to set a world record) do not get out too fast.  The first is to mark out every 200 meters for the first five or ten kilometers of the course, and employ several rabbits whose only job is to ensure that the pace for the first few miles of the race is absolutely perfect. 

The other way to ensure proper pacing is to offer strong monetary incentives to a second group of rabbits (or the actual competitors) to stay on pace for a long period during the race; e.g. for every kilometer run within three seconds of the goal pace, all the rabbits get an extra $500, with an extra compounding bonus for consecutive kilometers on pace. 

As anyone with a GPS watch has discovered, most people end up running longer than the prescribed 42.195 kilometers in a marathon.  Courses are laid out such that the shortest possible route—cutting all tangents perfectly, hugging the curb along corners, etc.—covers exactly the specified distance, plus a short course prevention factor of one meter per kilometer.  Again, two solutions spring to mind.

First, race organizers can paint a brightly-colored line that follows the shortest possible race route, so the athletes can just follow the line instead of worrying about plotting the optimal tangent to take.

Second, race organizers could petition the IAAF for an exemption to the short course prevention factor (which is 42 extra meters in a marathon!) and measure the course using surveying equipment or steel measuring tape.  Typical marathon courses are biked out with a device called a Jones Counter, and the one-part-per-thousand extra distance is factored in to take into account wobbling of the bicycle wheel, air loss from the bike tires, and other small errors.  The IAAF would still likely require some type of short course prevention factor, but with modern surveying equipment or steel tape, this does not have add up to an extra seven or eight seconds of running.

It goes without saying that attracting top talent takes a tremendous amount of money.  On top of that, to provide increased motivation for top performances, you'd need a prize money structure that incentivizes running a hard pace throughout the race.

As mentioned earlier, several top runners will need to be recruited specifically to serve as rabbits.  Pacing 25 km of a 2:02:00 marathon would probably require a runner in 2:05:00 shape.  Lower-caliber runners (say, 2:08 or 2:10 types) can be recruited to ensure that the early pace is spot-on, but it may not be possible to lure a 2:05-type runner away from other opportunities to serve as a rabbit.  Instead, prize incentives could be structured to encourage runners to act as rabbits for each other—bonuses to every runner who reaches 20, 25, 30, and 35 km on pace to best the previous world record, for example. 

To increase appeal for American audiences, 2:10-2:12 American marathoners can be recruited to serve as early pacers for the elite field.  Event coverage can focus on the human-interest aspect of the struggling, unheralded American pro trying to earn as much money as possible by ensuring a correct pace.  Additionally, if there are any adequately-prepared American marathoners, a "chase pack" could be organized with the goal of breaking the American record.

None of this will be cheap.  Major marathons fund their operations chiefly through entry fees from general participants and corporate sponsorships. Another reason to hold a general participation 10k race in conjunction with a professional world record attempt is revenue—the 10k race can help fund the marathon, and, exempting appearance fees, athlete stipends, and the prize purse, the only increase in cost on the margin is keeping roads closed off for longer. 

However, there's another revenue stream that's ignored at running events: betting.  Bookkeeping is a huge driver of revenue at the Kentucky Derby, which raked in $23 million from on-track wagering alone at the 2015 event.  For this to be a successful source of revenue, race promoters would have to put considerable effort into profiling and introducing the various contenders for the race.

Major marathons have yet to break through the general apathy over "nameless Africans" winning races—the promotion before and during the event needs to highlight the individual stories and athletic histories of each top champion who is invited to compete.

Relying on betting as a source of revenue puts some tight restrictions on where you could actually hold such a race.  In the United States, most states forbid gambling, so the list of possible locations is already fairly short.

In addition, there are a number of other amenities the ideal location needs: it must be at sea level or very close (the entire state of Nevada is out); it must be in a city big enough to host such an event (most Indian reservations are out); it must be in a region which has acceptable weather for a marathon world record at some point in the year (Louisiana is out); and the roads must be very flat. 

Other attractive characteristics would be a large pool of potential participants in the community 10k race, venues for hosting after-parties for participants to watch the professional race unfold, hotels equipped to host professional runners for at least a week beforehand, infrastructure for media broadcasting, and proximity to a major international airport.

By strange geographic, political, and climatological coincidence, only one American city fulfills all of these characteristics: Atlantic City, New Jersey.  Absecon Island, obviously, sits at sea level, and it is essentially dead-flat.  The nightly low dips into the perfect pro-marathon temperature range in the early and late winter.  Atlantic City plays host to a huge number of hotels and casinos, so infrastructure is no problem.  It's also an hour's drive from Philadelphia and two hours from New York City.  

Atlantic City is most famous for the boardwalk, but the  race itself should not run along it.  It could be uneven, and the springiness of the boardwalk might slow down the runners.  Theoretically, it's possible to design a surface such that its rebound properties speed you up slightly compared to a super-stiff surface like concrete or asphalt, but it's fiendishly difficult—it depends both on the speed you're running, and the individual stiffness of your legs, which is not the same, even for, say, all 5'11, 135 lb runners moving at 4:45 mile pace.  Further, this might set a bad precedent—future marathons might just be run on a track, or a track-like surface poured on top of a road.  The IAAF would probably rule that record-eligible marathons must be run primarily on asphalt or concrete anyways.

A lot of followers of the sport are fatalistic about the prospects of any non-Olympic running event getting national media attention.  The reasons have been expounded many times: marathons are too long to be interesting; casual runners don't care about international elites; there aren't enough fans of the sport to get prime-time coverage.

There is an excellent counterpoint to all of these arguments—a video game called Dota 2.  Unless you're a hardcore computer nerd, you've probably never even heard of it.  Yet it boasts millions of players worldwide, and Valve Software, the game's publisher, hosts a yearly tournament called The International which, in 2015, had an $18.4 million prize pool.  Of this, $1.6 million was contributed directly by Valve Software; the rest was raised through sales of a virtual VIP ticket-like item called a Compendium.  Though the event was broadcast for free online, fans who purchased a Compendium had access to exclusive information on the teams competing in the event and received a virtual goodie-bag of in-game prizes for supporting the tournament.  These in-game prizes help foster a massively profitable virtual economy, a large portion of which is driven by (you guessed it) betting on the outcome of professional matches.

You've probably never heard of Dota 2, but its prize pools and production value puts professional marathoning to shame.

Each ticket was $10, of which $2.50 went directly to the prize pool.  The rest, of course, went into Valve software's coffers—the company raked in over sixty million dollars in revenue from Compendium sales alone.

Dota 2 flies in the face of every argument about a sport's popularity.  The games are unpredictable—sometimes they're over in less than twenty minutes, while others can stretch to over two hours.  Often the games have long stretches of torpor, with opposing sides both too timid to make a bid for victory.  A tournament's grand final, which is usually a best-of-five match, can take all day.  Most of the best teams and top players are from eastern Europe and China, and many don't even speak English.  On top of that, the game is outrageously complex and infamously difficult to learn.  Yet despite all this, millions of viewers worldwide tuned in online to watch the grand finals in 2015. 

Compare these numbers to marathoning—according to Running USA, 541,000  people finished a marathon in the US in 2013.  From statistics on SteamSpy, around eight million people play Dota 2 at least once every two weeks worldwide.  Some six percent of these players are from the United States—this works out to an American player pool of 480,000 people. 

So, there are about the same number of Americans who play Dota 2 and who finish a marathon in a given year.  Though these numbers are similar, keep in mind, Dota 2 doesn't have nearly the pop-culture recognition as marathoning.  Everyone knows what a marathon is, but if you tell someone you're going to spend the weekend watching the Dota 2 Shanghai Major, they'll have a blank look on their face.  So why is it that Valve Software can put up $18 million for a video game nobody's heard of, whereas not a single World Marathon Major can scrape together even one million dollars for a prize purse?

The bungled marketing history of professional running is a topic for another day, but it's safe to say that Valve software's promotion of The International (and Dota 2 in general) is a case study in marketing genius.  It's a combination of capturing a huge proportion of the potential market of fans, integrating "core" sponsors (e.g. computer mouse manufacturers) as well as "lifestyle" sponsors (e.g. energy drink companies).  If you have some time, take a gander at the production value from the coverage of this past year's event—it's unbelievable for such an unknown sport.  When you compare this to the lackluster production value and commentating talent at most marathons and track and field meets, you start to realize the potential that's being squandered.

Putting on an event like this will still cost an enormous amount of money up-front, but I'm convinced that, under proper management, it can be a profitable endeavor—one which could be repeated every year, much like Valve Software's International tournament.

After considering the current shortcomings of professional marathoning, the path forward for new horizons at 26.2 miles should be clear: host a specially-designed event that takes place after 6 p.m. in cool weather on a very flat, loop-style course with minimal turns.  Minimize running extra distance by painting a clearly-visible optimal running line along the course, and petitioning the IAAF to use steel tape or precision surveying equipment to reduce the short course prevention factor.  Provide exquisitely controlled pacing for the first 5-10 km, and devise a generous prize money and cash bonus system to incentivize a fast pace throughout the race.  Gain broad appeal by hosting the event in conjunction with a mass-participation 10 km road race earlier in the evening, and get casual runners engaged in the event with high-quality production value, media coverage, and promotion.  Eliminate jet lag and travel fatigue by flying in the participating runners at least a week beforehand.  Finally, fund the event through a combination of corporate sponsorship, betting, revenue from the community race, and advertising during the media broadcasts.

Addressing all of these shortcomings puts some tight constraints on possible locations.  If we confine our options to the United States (because hey, what's wrong with a little patriotism?), by far the best candidate city is Atlantic City, New Jersey. 

I'm certain that an event laid out as I've described could result in at least a 2:01:xx marathon for men, if the best runners in the world show up in peak fitness and run under optimal weather conditions.

Now all I need is to find a well-connected politician or billionaire with ties to Atlantic City who's looking for something to do.  Hmm.  Any ideas?