Monday, June 10, 2013

Geoguessr


Someone referenced “cheating,” and I wondered how that is possible.  Apparently some folks think opening up other tabs to research what you discover is not quite fair.

I thought this was idiosyncratic, but checking the web and comments sections, this is the more-common opinion.  You can browse around a bit and read signs, but not google them.  The point of the game is the guess, based on your previous knowledge and what you can deduce.

I can’t say I agree.  I see the attraction of that style, as it is much quicker, and has that seat-of-the-pants, I-just-happened-to-know aspect that’s fun.  I’ll play a few rounds in that style to see if it seems better to me.  Yet my initial inclination is that the quick version is inferior.  It appears to reward skill for most of its playing, but at the last moment becomes a game of luck. Many cities of 50,000 in the US look pretty much the same.  Unless you are following a road out of town to see the terrain, it’s pretty much a guess on Bozeman vs. Nashua.  And if you’re in that deep, I don’t see the difference in using another tab.  Dead flat, little vegetation, snow covered, two-lane highway with signs in English.  It could be Arizona or Saskatchewan.  Probably something in Australia as well. So you make a good guess and are off by 2000 km.  In Europe, that would be a terrible guess, but it’s scored the same. Brazil is a big place, and villages far apart look darn similar, even if you were clever enough to recognise that the sign is in Portuguese, not Spanish. Korean downtowns look similar – unless, I suppose, you have traveled to many of them.  Ditto Slavic cities.  Or rural areas in the Balkans and a lot of Eastern Europe.  Do you read Cyrillic characters?  Or is that supposed to just tell you “guess something Slavic?”  When each site has something unique that you might know that separates it from similar items, whether vegetation, or architecture, or some geographical feature, that’s skill.  When a lot of sites look pretty much the same, that’s luck. 

The scoring, BTW, is not linear.  It falls off rapidly from about 6487 (1 meter off), so that being 300 km off is not 100x worse than being 3km off, but only about 10x worse. And 3000km off is not 10x worse than that, but 3x.  That alone suggests that precision, rather than inspired guessing, mattered to the designer.  Not that the designer’s word is necessarily final in such matters – but it should be accorded some respect. My low score for a site is 74 BTW: 16000km off.  Australia looking like North America in many ways will do that.

News articles kept suggesting that high-scoring is not the real point, but the adventure and guessing part is.

Then why keep score at all?  People say that about kids and sports, and it’s a perfectly valid way teach kids physical activity, appreciation for the game, etc.  What matters about golf is the exercise, and being with friends, and the beautiful surroundings, etc.  But then don’t keep score.  Once you keep score, score matters, to some people at least.  If it doesn’t matter to you, fine.  You can play basketball or golf without keeping score if you like.  But don’t say that those who do keep careful score aren’t doing it right somehow.

Second point.  When bapping around the terrain looking for clues, some signs are intentionally obscured and some are not.  As I have been using a separate google tab and can check back on some of these, I know what the trend is on those.  The smudged signs are more unusual items that a google would narrow down quickly.  “Wilson’s Auto” or “Park St” or “Memorial Field” are not obscured. “Grebb’s Pizza,” “Shearwater St,” and “Briggiano’s Field” are. You would only know those by luck, but you could get within 100 meters ridiculously fast.  Sometimes route numbers are visible, sometimes smudged. Therefore, we conclude that the designer expected googling, but took pains to make sure it wasn’t too easy. 

And it isn’t easier.  Sometimes even when you get a clue that gets you in the ballpark, something crosses you up.  You can find an equipment trailer from a marine construction company in British Columbia, but the construction could be far distant.  It was.  Northwest Territories. I knew immediately that one place was Appalachian, probably West Virginia.  A half mile up the road was a Kingdom Hall, and just beyond that a non-chain motel.  You’d think you could zero the place in from that alone.  But all my googling showed no motels within a mile of any rural Kingdom Halls in WV.  I was off by 40km when I gave up and guessed. (The motel is now closed, that’s why.)

I think it’s just playing at a different level of intensity, with luck more of a factor in one.

Hmm. 16,722 in 13 minutes.  As opposed to 32,000, usually taking an hour.  Well that’s fun.  But this one was weird.  CA- Oregon border, Oregon, British Columbia, Alberta, then Castille, Spain.  Those first four were pretty tight.  That doesn’t happen much.

19,430 in 19 minutes.  Got a couple of very lucky breaks (border of Maine and NB, for example).  Maybe this is better.

Or Speed Geoguesser.  15 minutes max.  Should be fun.

2 comments:

GraniteDad said...

I was going on pure guess- rotating the camera but not moving at all. That's very intense but also frustrating, as it's impossible to be really closer than a state in the US, and lucky if you get the right region in Europe. Though only being off by 130 Km in Romania was pretty awesome. I think I'll have to try at least moving around within the images.

Assistant Village Idiot said...

Speed Geoguessr is the ticket. Google if you want, but that takes time and you'd better be good.

I got a 61 on one entry, and under 1000 on a few others. But it's very satisfying to get within a few hundred kilometers when you only have 3 minutes.

I had a Romania - got it from the white tree-bases and guessed the mountains near Beius. 112 km off.