Comment by tofof | Hacker News Reader

Comment by tofof | Hacker News Reader

Unimpressed by this game. The accuracy on the scale is off by as much as 25%, as can be seen here: https://i.imgur.com/NOvqCnV.png. As the only information you're given other than a new photo is the distance, this is a serious problem.

What's worse, the 'correct' location radius doesn't even include the locations where all the photos were taken!

SPOILER FOR GAME #60:

Game #60 has a clue that shows YOU ARE HERE: https://i.imgur.com/98OXlal.png which is exactly here (google maps [1]) https://i.imgur.com/KdN4OCh.png which corresponds exactly to my guess location https://i.imgur.com/eCy97r6.png. However, they claim that the correct location is here: https://i.imgur.com/cSasI3g.png or https://i.imgur.com/ChuisIx.jpeg (google [2]), which is actually 9.5 miles away and outside their 5 mile radius, so it scored me as missing by 4.5 miles.

It's quite simply unacceptable to claim that all the photos are of the same place and then have the exact location within the photo not be within the accepted answer radius. Geoguesser, for example, REQUIRES you to identify the exact location within ~150 meters (in the worldwide mode) and they are actually correct with that degree of precision. This error of 9.5 miles is 15,000 meters away.

1: https://www.google.com/maps/@43.3419792,-122.7415299,14.62z?...

2: https://www.google.com/maps/@43.2970892,-122.5563029,1891m/d...

Yeah they should not be using a Mercator projection if a key part of the game is triangulation AND there is a ruler shown on screen.

Yeah my first guess on #74 was 7.7 miles away (!) so I thought for sure I would get it. But it turns out that the flag location is actually 14 miles away according to Google Maps, and the multiple photos "of the same place" are over 6 miles apart. So I kept getting weird numbers for my guesses that didn't line up with the geography.

Man AND the ruler isnt the same size every time.

I kept getting wrong answer after wrong answer despite the fact that I was CONVINCED I was triangulating properly.....

But no! The size of the ruler kept changing!! I was using a constant 80px for it, and it was changing by as much as 30px as you zoomed in and out.

What a waste of time.

However I did get within 7mi once despite this terrible UX, because I saw a pic and was like "goddamn if that's not in non-Chicago Illinois" so I was within 200mi on my first guess and knew how to constrain my further guesses.

Maybe I’m misunderstanding you, but that’s exactly how I would expect it to work and how it does work with Apple Maps and Google Maps. You can zoom at different increments than the scale label shows. Without changing the ruler size, how could this work? Am I misunderstanding you?

What I mean is like: I expected the different zooms to be specifically normalized so that at the most zoomed-out, 80px = 800mi, then 80px = 500mi, then 80px = 300mi, then 80px = 200mi, then 80px = 50mi, etc. But instead it was like.... 90px = 800mi, and 80px = 500mi, and 60px = 50mi, etc. So not only do you have to account for the changed scale, but you also have to measure the scale ruler at every increment.

Ah, you want more distinct zoom levels. On my mobile device, for example, the ruler for, for example, 1,000 miles can be different lengths depending on if I am slightly more or less zoomed in (via pinch). Which might be related to the choice of projection, because even just moving around the map changes the ruler length for me, for the same labeled distance.

Hmm for me I did have very distinct zoom levels here; I was only able to scroll in one tick or out one tick with my mouse wheel, it wasn't freeform. That's why I expected these quanta with a constant 80px-long ruler. Maybe on mobile it was a different experience.

Reply on news.ycombinator.com