This thread was originally titled, "Incredible development in the $612,000 Transcon Goodge run, currently ongoing" but the new title is more descriptive. The description of the run is here.
I think it would be better if people just stopped replying to the trolling of Sneakers. Ignore him and he'll go away. Nobody takes him seriously.
You wrote this, which is the correct thing to do with trolls, but then immediately after you replied to one of his posts, which then invited that douche canoe to post yet again. Wtf dude? Lead by example.
FFS, why should it require computer skills of ANY level to be able to verify that what this guy is doing is legit? If he were legit, he would WANT his efforts to be easily verifiable, and would respond instantly to all criticisms and requests for back up verification, such as the Garmin Connect links.
Liken it to an athlete suspected of doping. Clean athletes use any and all means available to remove doubts and allay concerns, starting with the simplest step of always being available for testing when contacted. They don't carry out their own private testing, fail to show up for official testing when asked, then require that their fans know how to download obscure background files on their private testing protocols and results in order to allay concerns about their lack of compliance with standard testing requirements. This is precisely what you're asking people to do on this fraud's behalf.
But you're a troll and already know how ridiculous all of this sounds. The only question is: What exactly is in it for you to persist?
Sneakers, why do your plots show a cadence of 0 corresponding to a speed of 3-3.5mph?
Mostly because you are can't zoom in enough. There are 39465 data points used on those graphs that are only about 1000 pixels wide for the plot area. Meaning each pixel in the graph is representing about 39.5 data points corresponding to 39.5 seconds of real life.
Creating a stride length histogram that calculates the stride length using both the speed and cadence makes it very easy to see that there are relatively few data points where PJ is moving but cadence is zero. In the histogram below, that would be the very far right side at the 25+ feet bar where I chose to end the histogram. It is expected there would be a small number of data points like this simply because watches can fail to detect a step if he is doing something with the arm like looking at a GPS, drinking from a bottle, using a towel, etc... Or it could even occur when he is actually stopped but there is GPS signal jitter and he is moving his arm.
Paul's stride length for 2024-03-11 was mostly in the 4-7 feet range (distance covered by both legs together). Nothing unusual about this day.
2024-03-11 Paul Johnson Stride Length Histogram (distance covered both legs together)
That’s a good comparison. If you’re doing nothing wrong, then you have nothing to hide. More transparency and more data will only prove that you’re on the up and up
That’s a good comparison. If you’re doing nothing wrong, then you have nothing to hide. More transparency and more data will only prove that you’re on the up and up
I am sure you would like to post your entire web browsing history for this year for us to check your transparency
That’s a good comparison. If you’re doing nothing wrong, then you have nothing to hide. More transparency and more data will only prove that you’re on the up and up
I am sure you would like to post your entire web browsing history for this year for us to check your transparency
It’s really odd to attack people who are looking for proof. For example, Guinness, who verify world records, don’t seem to do that. They verify if the person has done what they claim to have done. They don’t go to everyone else “well let’s see you do it then” or “let’s see your browser history then” lmao. An odd straw man argument to cling to. But the burden of proof lies with the person making the audacious claim. If you’ve really done the thing it won’t be hard to prove it.
If I could be bothered, I'd look into it. However, it's irrelevant as that data wouldn't be available for a wider audience. My Garmin data is publicly available. There's nothing to hide there. Wonder why Paul isn't releasing his?
If I could be bothered, I'd look into it. However, it's irrelevant as that data wouldn't be available for a wider audience. My Garmin data is publicly available. There's nothing to hide there. Wonder why Paul isn't releasing his?
You've spent a lot more time in this thread posting that it actually took for me to figure out how to do the data downloads. It's not that you can't be bothered but it is because you lack the skills to do what I have done.
Why do you think some of the elite women runners hide their Strava and Garmin accounts? They don't want you stalking them.
Download the GPX from Strava, download the heartrate+cadence data from Strava streams API (ex Morning_Run.json) and this will generate a csv with all the data; time, grade, distance, lat, lon, elevation, heartrate, cadence, speed, pace, etc.
Python peoples do double check it and extend as you please. I added a couple example alerts for things like pace over 12 mph.. and wowsers, unless there is a bug, there are points at over 50mph.
I had a look at some of the raw data and it actually seems pretty clean so far.
There is a Chrome extension called Strava Sauce which allows you to download someone else's .fit or .tcx, among other features. Then there are sites to convert that to .xlsx. Sneakers could have saved us some time on that front but anyways.
Going to keep poking around for the smoking gun, if anyone has suspicions about a certain mile or section so far let me know and I'll inspect.
Sneakers - wondering what methodology/rules you are using to smooth cadence? Thanks
This post was edited 1 minute after it was posted.
Download the GPX from Strava, download the heartrate+cadence data from Strava streams API (ex Morning_Run.json) and this will generate a csv with all the data; time, grade, distance, lat, lon, elevation, heartrate, cadence, speed, pace, etc.
Python peoples do double check it and extend as you please. I added a couple example alerts for things like pace over 12 mph.. and wowsers, unless there is a bug, there are points at over 50mph.
Good effort but generally you do not want to be using the speed data from Strava like that since Strava will calculate an instantaneous value from lat and lon if the watch itself doesn't record a speed field. And also, if the speed value is one record by the watch and PJ's data is at 1 second intervals, there will not be much smoothing of the GPS position errors which can cause momentary spikes. If you are seeing 50 mph, you are very likely doing something wrong or Strava is doing dumb calculations for the speed values.
You should instead be calculating the speed yourself from the lat and lon and apply several seconds of smoothing to avoid most of the GPS errors.
As far as speed goes, PJ's 2024-03-03 day is a bit suspect for me. The only day like it where the speeds are abnormal (but not ludicrous like your 50 mph). Was he cheating on that day? Maybe? Is there some other explanation? Possibly?
I had a look at some of the raw data and it actually seems pretty clean so far.
There is a Chrome extension called Strava Sauce which allows you to download someone else's .fit or .tcx, among other features. Then there are sites to convert that to .xlsx. Sneakers could have saved us some time on that front but anyways.
Going to keep poking around for the smoking gun, if anyone has suspicions about a certain mile or section so far let me know and I'll inspect.
Sneakers - wondering what methodology/rules you are using to smooth cadence? Thanks
For cadence, I output both unsmoothed and smoothed cadence values for my analysis from the downloads. For the smoothed cadence, I just use a simple 20 point moving average here since the data is at 1 second intervals.
For speed, do not use the values Strava provides because it essentially calculates instantaneous speed from the lat and long without any noise filtering. I'm using a Kalman filter on the lat and long coordinates with filter accuracy set to 50. Otherwise, you are going to get ludicrous speed momentary values of 50 mph like the other poster.
If you want to really dig in, the only day I would call suspect is 2024-03-03 because I can't explain some of the high speeds that occurred starting at 4 hours into the run. Lets see if what you get for the speed plot matches what I see.
I won't post again till completion - let's see what gives, and speak on the other side. It will be an intriguing denouement whatever happens.
As someone who has the utmost respect and appreciation for your commitment to integrity in ultra running, I will miss your daily reports. Reading your daily takes has been like watching Sherlock Holmes investigate a case. It’s become my favorite part of this site.
I hope others post updates, as I refuse to give this Paul Johnson fellow any views. His sudden metamorphosis from an in over his head rookie to elite level runner is highly suspicious.
Respectfully Sneakers.. You obviously didn't read the code and you don't seem to know the data Strava gives you when you download PJ's activity GPX or hit the streams API. I don't use speed data from Strava and speed is not in the GPX or the streams API json. The speed comes from my python code calculating distance between two sequential latitude longitude coordinates, from the GPX file. Look at the code, I used geopy.distance.geodesic() to get the distance and calculate speed from there. You can add smoothing if you want but I was just using this code to flag specific areas where it looks odd and then I look into those times/places more closely. Anyway I agree, I very well may be doing something wrong but I posted the code and so if I am.. y'all can just point out the bug and post a fix for it here.
