Another run across America underway - is it legit? Boots currently on the ground

I don't understand your post. I'm not defending WG. Neither am I calling Garmin or anyone idiots, I'm citing studies. And I don't need to say why data is dirty because I'm not calling it dirty, I'm pointing out that at least some data you label dirty is legitimate, but you're not addressing it.

Well... of the 430-odd K I observed around 400 were fine. Yes, there was an odd spell with Balenger, on one of the days. I think they threw that in to mess with me. Fact is there have been some 150 out of 175 days like that, who knows what's going on, but something is afoot.

Either:

It's a catastrophic tech fail and overwhelming coincidence that these guys have data fails like this to such prolific levels, at such specific times

or

they're doing something untoward.

Around 97% of the folk on here believe it's the latter, you the former - fine.

If I hadn't fought so hard on this, I wouldn't have brought it to such a large audience. I wouldn't have a scribe from Runners World commenting on GMTVs post. I wouldn't have statisticians from one of the world's top athletics universities, and a leading cardiologist, working on this. I wouldn't have had hours with one of the world's leading broadcasters today, or 2 hours in the last month with a streaming giant, or two more hours with a writer at a major ultra magazine; or people pointing out to me Balenger's identical modus operandi, that would only have been found due to this thread. I wouldn't have people screenshotting his Whoop to show glaring, shocking anomalies.

This is one of the larger threads in LR history because it raises a very important matter: social media influencers coming into the sport, turning pro, getting all the sponsors denied to far worthier and more deserving cases, raising huge funds on their Gofundmes and making outrageous, false statements like they're faster than a car, world record holders and best of British.

But you just dismiss all this and say, waste of effort and echo chamber.

Has anyone requested the direct watch data files for expert analysis, and if so, has the team provided a response?

Because you, Will Cockerell, are a fraud who claimed to have analyzed the entirety of all the data on Strava and you fabricated your "statistics". As proven in this thread, you didn't even download a single one of the data streams from Strava and you have no idea how to even download it. You pose as someone in authority when it is clear you have no skills at all in data analysis.

You asked earlier who I am. I'm your worst nightmare because I actually do data analysis and know what I'm talking about.

Ooo, the self-proclaimed super-genius Sneakers is back! His evil superpower is the ability to perform amazing feats of intelligence in utter secrecy, and then...not share those amazing feats with anyone! He writes brilliant software which proves that Goodge didn't cheat, and then brilliantly won't show us any actual results! Soooo impressive!!!

No you don't. You've been asked to provide even a screenshot of your scrawled "math" but you can't, because you're just an average runner with a moustache beating off in a rented trailer with a fake record painted on the side. You're full of sh...

Have you ever been in a pub and some bloke bores you rigid for 20 minutes with how he invented something notable but never got the credit, fame and wealth?

That bloke is exactly the same as you, down to the way no evidence is forthcoming.

😂😂😂😂😂😂😂😂😂😂😂😂😂

Pray tell how you think you have the intelligence to understand what I would write seeing how all you Cockerell shills turned to mumbling idiots when Ashley Paulson released her data files and couldn't even tell a fudge popsicle from a pile of poop.

This is truly hilarious. The momemt Sneakers On The Ground started to think he is Freddie Krueger!

Hey Sneakers, if you are so scary how come you are too scared to tell us who you are?

pupildidnotwould

Ooopsiedaisy. Read your post carefully and understand you have just admitted to lying.

Does this mean that we can expect to see all of the raw unadulterated watch data to exonerate William Goodge of accusations of possible data manipulation and to prove that he ran every step of this journey?

If you reign in your emotions a bit and read my post you'll see I'm actually giving you some advice. You are now trying to convince a wider public, and my advice is to think more about how present this. If you rest on heart rate data that argument will be demolished.

A) Whatever you say, it's not that reliable. All you're going on is your own anecdotal evidence and an unrepresentative, cherry picked study. Countless people have been on here to tell you their own data isn't reliable but you refuse to listen.

B) There are unusual patterns to WG's data, undeniably, but you undermine the credibility of your argument but overplaying it. I cited four occasions he recorded low HR data when running, all you have to say is you think they manipulated it in front of your eyes, but you have nothing to suggest how. Then you're very fond of saying that his heart rate doesn't follow the same pattern outside of these events. But he doesn't run at his ultra pace in training, like everyone else, so it's not relevant. Outside of an accepting audience this will be used to rubbish your findings.

Try seeing through a layman's eyes:

You were sceptical of WG's claims so you went to observe him. When you were there he covered the same distance as before in the same time. His heart rate recorded low while you watched him run. He was being filmed. You accused him of watch swapping, you no longer do, you just think something is up but you can't say what.

Can you not see that isn't very convincing, and that the way to convince someone isn't to say "well look at his mate's heart rate"?

You're basically defending his unusual patterns by saying as he's not doing multiday pace elsewhere that they're explicable.

Either this is is a tech fail, or it's not. Which is it? You simply refuse to commit to that.

And his patterns changed in all sorts of ways with me, that have been gone over loads.

And now you dismiss the link the RB.

Actually the general public are being surprisingly docile when we comment on media reports, we just get likes and comments saying 'I concur'

There isn't a study around (and I went through a huge number of them) showing anywhere near the number of data fails as Goodge and RB, and to anywhere near the same degree, even on equipment a half decade (or more) old.

Yes, it happened when Will C watched (notably day 26 - and around page 40 I was in complete agreement with you, if you recall), but the overall period of time in which this happened during that 5 day stretch was significantly less than it was prior to him leaving.

I also think its highly suspicious that they can seemingly correct the data when pressed, but 1) haven't bothered to do so the rest of the time, 2) haven't explained what's causing it publicly, despite the bad PR from it (if they made an instagram post explaining it, this entire 2300 post thread would be in shambles), and 3) if they (somehow) haven't figured out what's causing it, then at least being curious enough to figure it out (which shouldn't be too difficult).

I don't think your points are without merit. It's a bit of a sticky situation where WC has been a bit hyperbolic at times, but the weight of evidence - and the sheer weight of coincidences - combined with the conduct of WG's team (including now falsely claiming a record they know they're falsely claiming), is pretty substantial.

There is some wisdom here. I'll try to add some. A bit of moderation, humility, and a thick skin would go a long way. There are people advocating for two sides here, and they can learn from each other, if they don't exaggerate each other's claims and if they treat each other with respect.

First, Sneakers has done some real work on the data. He has chosen some metrics, provided the link to the data source (the only to do so, if I recall correctly), described his method, and provided his results in an understandable format. Not only is this far more than most others appear to have done with the data, but he has been pretty transparent about it. For people to question him because he hasn't provided all his code is ridiculous. I can understand a digging in of the heels by Sneakers (but not the arrogance). Spot check his work, and if you find something amiss, assess together with him whether it is you or he that has made an error. Until that has been done, I see no basis for questioning the work he put in (which, granted, was largely automated). For people to question the legitimacy of a stride length calculation, given cadence and ground speed data, indicates lack of understanding, plain and simple: what is simpler than dividing two numbers, which Sneakers said was done [correctly] with smoothing. I've observed in the past that there seem to be many on LRC who are capable of the data analysis Sneakers did, and who could easily check whether it is correct. Approaches to noise reduction might differ a little, but the time to discuss those differences is after an attempt has been made to reproduce some of his results.

On the flip side, Sneakers has rested on his chosen metrics, as if they settled the case and there were no need to look at anything any more. They do not. They are his chosen metrics.

willvlc's strings of numbers, with no units provided, manually typed in, are hard to understand to those that aren't on his wavelength. Providing those numbers without context (as a generalisation) as to uphill, downhill, etc is not helpful. Accompanying them with less-than-objective language does not help those of us who trust data, and reeks of fanatacism. The legitimate points he might have are obscured with cyptic presentation and prejudicial verbiage. I have found willvlc generally presenting conclusions based on his own judgment (which might be very good), but in a way that doesn't communicate the alleged obviousness of the problems easily to others.

I've found the anecdotal reports unclear and therefore unpersuasive. 'I was there there' said WG's heart rate is low -- did he tell you, or show you; and if the latter, was it demonstrably current? What was the principle of operation of the monitor, and was it demonstrably reading his own heart rate? Etc. At an earlier time, willvlc allegedly watched and yet seemed not to be able to watch much of the time. How did that go? It wasn't documented well, so these things aren't clear to observers of this topic.

Having said all that, willvlc and others have at times pointed out some legitimately suspect-looking matters in the data. I would have expected the likes of Sneakers to take a closer look at the data in that respect, instead of depending solely on his broad averages. He has the acumen to work the data and seems to understand the issues involved in data collection. Is he willing to put his own conclusions to the test? But it shouldn't depend on Sneakers. Others should be able to verify or dispute the claims and be able to present the case in a readily understandable format that shows the anomalies.

On the matter of whether heart rate data can be depended upon, I submit that if there is a heart rate problem, it should show in the remainder of the data. Heart rate will reflect effort (broadly speaking, and with a bit of lag for physiological response to change in effort), +/- cardiac drift, so based on the work being done, a proxy for HR can be derived. I have not done a lot of work on the data; I don't have the time. But I looked at one day's run, and the data looked surprisingly plausible to me. I saw no indication of a tech failure in the HR data. I did, however, observe what seemed to be different signatures in different parts of the run, and not only in the HR. The differences are enough to make me suspect foul play, but not enough that people couldn't argue them visually. And I haven't been able to quantify them, so I'm not presenting anything conclusive about it. But others might be able to.

LRC has many capable people with the skills to tease this kind of stuff out of the data and present it for others to look at in an easy-to-understand format. Why is it that not one person has done this? We expect a standard for burden of proof on someone doing a transcon. Anecdotal evidence doesn't fly. Well, there's a standard for debunking based on data as well, and you have to clearly show it from the data. I don't think willvlc has met that standard. He might need help in the communication department: fair enough. Someone needs to turn his strings of incomprehensible numbers into comprehensible graphs that compare apples to apples and demonstrably exclude confounding variables like pace, grade, conditions, and, for historical data, development. Those graphs need to be accompanied by sensible commentary that understands the potential objections and deals with them. Then there will be a case that anyone can look at and understand. Data speaks, not the personal judgment and anecdotal evidence of someone on a vendetta. If you want the world generally to believe a claim of fraudulent behaviour, it needs to be backed up properly in a way that the world understands.

Thanks for reading.

Thanks. As you know I have many problems with the run, there is no vendetta or fanaticism, and to say I haven't been clear, when I have persuaded so many to the problem, is untrue. But in short, and crisply:

1) WG and RB file their HR data at 110 for these runs 90% of the time, over 4 years, with up to six different watches, over some 14,5000k. The rest of the world files at 125-175, with clean data 98-99% of the time. No one can say that this is irrelevant.

2) You say you have looked at just one run. It is brutal to argue against me so harshly having looked into this so little. There have been 194 runs. 75+16+48+55, and that's not counting another couple of v strange ones of RB. Perhaps you could care to look at a couple more? How about WG Transcon day 8?

There are up to a dozen other keys problems like the failures of Whoops to match with Strava or present credible calories burnt; his incredible splits and getting around the second half of the course in the sixth fastest of all time, with a 5 day negative split.

And now WG and his business associates are claiming to all and sundry the British record when he didn't get it. When we point this out, we get deleted and blocked. Interestingly, Andi Rivett who claims the fake record for Jogle, has also accused me of a vendetta. It seems to be the go to response in cases like these. Ergo, I also have a vendetta against Rosie Ruiz, Mark Vaz and Rob Young.

But I am content to let the rest of the world take over now. The matter is now with a key university who need a week to ten days to crunch the numbers, and other major news and media outlets. It may well be that LR have indeed taken this as far as it can go. We are not the outside world, and I won over around 95% of this room, it's enough.

Until we meet again!

Oh I do. Trust me. All we ask, oh super smart ass is that you just show us some copy... Whether we understand it or not isn't the issue, its the fact that your research doesn't exist and will not exist because you're making it up.

I argue that this is largely false. The whole "optical HR monitors are inaccurate" is largely a thing of the past. Do I want my cardiologist using a Garmin? No. Do artifacts ever appear in my HR data? Yes, extremely rarely, and it's usually picking up cadence. There's not many studies about this, so saying Will's is "cherry picked" seems like a bit of a stretch. Here's another study:

nature.com

Investigating sources of inaccuracy in wearable optical heart rate sensors - npj Digital Medicine

npj Digital Medicine - Investigating sources of inaccuracy in wearable optical heart rate sensors

Open link

. I skimmed it, so feel free to correct me if I missed crucial pieces, but the takeaway seems to be "Our conclusions indicate that different wearables are all reasonably accurate at resting and prolonged elevated heart rate, but that differences exist between devices in responding to changes in activity." Also interesting that they noted "...the measurements from the research-grade wearables were less accurate than measurements from the consumer-grade wearables." Many years ago, before I even had a GPS watch, I remember people talking about how bad the HR was. Now if I go on Strava and pick any of my friend's activities and look at the HR, it's extremely rare I spot an anomaly. Yes, I often look at this stuff out of boredom.

After 4 years of a computer science education and a few years working in research, never once have I heard someone say it's ridiculous to provide source code. I also provided links to the Strava streams, the same data that Sneakers claims we are not smart enough to access. I would love to do some real analysis, not an Excel histogram like Sneakers, but I'm too many years removed from this. It would take me a whole-@ss weekend because I suck at writing Python now. I proposed better methods for analyzing HR, but Sneakers decided it wasn't worth taking a look at it.

Agree

Agree. This is why I would like to see a proper analysis and write up done. Someone cut me a check and I'll do it lol. Getting the data organized and processed is one thing. Putting it into a convincing and legible format is a whole other thing. I'm way better at the latter these days.

Agree and disagree. After I read Will's posts for the 100th time, I was convinced. But they're not convincing anyone fast. And they're not convincing anyone with a short attention span. The use of language to describe what was going on was, erm, sub optimal at best.

Write me a check and I'll make you a pretty web page with responsive graphs that tells a story :) Otherwise, I have work to do. I'm hoping someone who is much faster at data analysis does this. Like I said, it would take me a weekend, maybe more, to get through it. A seasoned pro could bang this out in a few hours.