There are a lot of ways to describe the path that new technology takes towards being adopted.
There's the typical 'life cycle' chart, probably a product of business schools, the population (market) divided into (terminology may differ) 'innovators' (nerds), 'early adopters' (wannabe nerds), 'the majority' (normies), and 'laggards' (a stand-up comic's parents).
Then there's Douglas Adams' three rules: 1) Anything that is in the world when you're born is normal and ordinary and is just a natural part of the way the world works 2) Anything that's invented between when you're fifteen and thirty-five is new and exciting and revolutionary and you can probably get a career in it 3) Anything invented after you're thirty-five is against the natural order of things.
And then there's Google Glass.
I looked it up recently and Google Glass, or the big Google Glass hullabaloo, was 2014. Nearly a decade ago. If you said 'Google Glass' to Ansu Fati - aged 19, born late 2002 - he'd probably wonder why you wanted to do an internet search for such a common material and that, besides, he'd sooner look it up on TikTok anyway.
Google Glass may well make a resurgence in time, but it never even got to the early adopter phase of the adoption chart, never got people of Ansu Fati's (current) age thinking that this was something new and exciting and that they could probably get a career in it. It was a little too disturbing, a little too complicated and expensive, a little too 'well what do I do with it?'.
Football analytics has gone from strength to strength since 2014. If Google Glass had had the rise that football analytics had had since then you'd all be reading this a quarter-inch from your left eye while watching TV with the other. Even the laggards know what xG is now.
But in all those years, y'know something that all this data and all these genius minds haven't dealt with much yet? 3D.
For all this time, the rocket ship of statistical soccering has grappled, in large part, with just the X and Y planes of existence. (I say 'just the X and Y' instead of 'just two planes' because this is the kind of newsletter whose readers will write in to point out that football analytics has long dealt with time, itself a dimension — yes, I am familiar with the 'normal chess is 3D chess, 3D chess is 4D chess' argument).
3D data is less important in football than in some other sports (where it's already been worked with) given that most of the action happens at, or in the neighbourhood of, ground-level. It's also a sport of not just low scoring (making it harder to work on repeatable scoring mechanics), but of low 'opportunity to create scoring chances'. Sports get war metaphors a lot, but football isn't a series of skirmishes, it's career planning in the sixth-form guidance counsellor's office. If you lay the right tactical foundations and keep good structure in possession maybe you can land an entry-level job interview at Natwest.
So, unlike basketball, where 'body pose' software was used in 2017 to analyse NBA player three-point shooting motions (Ansu Fati was not yet fifteen), 3D skeletal data in football has been, to Get Goalside's knowledge, scarce.
Is it because, like Google Glass, it's a little complicated and expensive and honestly a little silly-looking? Maybe.
Unlike Google Glass, though, it's getting less and less 'well what do I do with it'. A paper earlier this year used body pose-adjacent data-gathering to analyse Leonardo Spinazzola's Achilles tendon rupture. At this year's MIT Sloan Sports Analytics conference, another paper took goalkeeper body pose data from one-v-one situations to analyse save technique decisions. (I wrote a 'Research in Focus' summary of that goalkeeper paper here).
That second paper is an example of the kind of isolated technique that lends sports like baseball and cricket to such a wealth of quantitative analysis - and, as it happens, use of this kind of data. Driveline Baseball has its own biomechanics lab to use data and luminous skeletons to improve performance. Sky Sports debuted some stick figure analysis in their England-South Africa Test match coverage earlier this week.
In the old days, shortly after Moneyball (the movie) came out, people said that the reason we didn't have an Oakland As or a Billy Beane in football was that baseball was just a more quant-y sport. It was all about numbers, and football wasn't. Some said that data couldn't work in football - or at least not anywhere near as significantly - for that reason. They were wrong. (We now have several Billy Beanes in football, as well as the actual, first-edition Billy Beane).
It would probably be wrong, then, to sit here and say that football - unlike basketball, baseball, and cricket - simply isn't a sport where body pose data would prove to be useful.
Goalkeeping (as readers of the aforementioned paper or attendees of previous StatsBomb conferences will know) is an area ripe for its use. The current trend in set-piece focus also feels like it would meld well with skelly-data analysis.
Other discrete skills are probably a bit more difficult to capture and train in settings where you can obtain high-quality data, but surely not impossible. A step removed from set-piece shooting would be the 'Coutinho' or 'Robben' shots: wide forwards cutting inside to the shoulder of the box, where there is usually some space. These situations seem more replicable than many types of shots, and offer relatively few options on the shot itself. I'd be intrigued at the idea of using this approach to study players' control of bobbling passes too, as a frequent part of the game that seems potentially reproducible in lab-like settings.
To use the Douglas Adams criteria, people like me were born into a world of event data - Player A tackled in (X, Y) location - it feels normal and ordinary and just the way the world works. Body pose data is new and exciting. Potentially revolutionary. Is it in, or approaching, a place where you could make a career out of it?