Comment by psn | Hacker News Reader

Comment by psn | Hacker News Reader

I found this a pretty interesting read. hackerish reasons: - The scale of the fraud. This guy is making up a race on the internet. complete with other runners, their profiles and so on. Its the sting come to life.

- The data based approach to cheating detection. Random runners can look up, in seconds, stats for any other runner. Not just their times and placings, but checkpoint by checkpoint timings. Its then easy to profile the normal runner and then the cheat.

- The photo aspect. disqualifing a guy because he didn't appear in any photos shows how far we've come in having 100s of photos of every event.

- The lack of on-the-ground cheating detection. This guy presumably left and entered the route of the race, with no one noticing. Runners noticed themselves going from position X to X-1, but not ll of them complained.

- The issues of internet anonimity. We see yet another internet comminity with pseudonyms. This doesn't prevent it being a commumity. it doesn't prevent people spotting strawman accounts.

and less hackerish reasons: - The lack of clear motivation. Why did he do it?

- the weird overreach. Why did he aim for a high ranking? If he'd come 50th no one would have cared. Why did he talk to someone from the new yorker? Why did he make up a whole event?

People do step off courses--for restroom stops, to chat with friends, to have blisters tended--and then back on. I once stepped out of the Marine Corps Marathon about the 10 mile mark to chat with some friends who were near the Georgetown end of Key Bridge, for I knew already that my time would be nothing special. When I pushed back in, there were some spectators who supposed I was cheating, and why not--there is nothing so memorable in my appearance that they should have noticed me leaving the course.

"- the weird overreach. Why did he aim for a high ranking? If he'd come 50th no one would have cared. Why did he talk to someone from the new yorker? Why did he make up a whole event?"

This is what surprised me about him. If he had more respectable times, he could've run the scheme for much longer, and avoided the scrutiny of the community. Even the author at the end said his sub 4 hours times were believable and gave him credit for those.

As always, his greed got the best of him.

> "The data based approach to cheating detection."

I've been really impressed by this recently. Even in areas where it doesn't seem like it should be hard to plausibly fake, it turns out there's a lot of subtlety that might be missed.

Here, it was the checkpoint-by-checkpoint splits and the photographs. In a fake Minecraft game [0] it's the number of blocks that have been mined versus the materials available for building tools (if sticks are rare, you're not going to waste them making pickaxes to strip mine plain stone) and the number of structures that have been built (millions of blocks got mined, but not used?) I've thought about building a "wall transparency" detector for another video game that would track a player's orientation in relation to players behind a wall from them, profile the average player, and then show that certain players are able to track others through walls.

There is an incredible amount of technical knowledge even in seemingly mundane pursuits like running. That's what makes it possible to spot cheats even without spotting the exact cheating mechanism.

[0] http://news.ycombinator.com/item?id=4413499

Reply on news.ycombinator.com