iNatGuessr - Can you guess the location from the observations?

[simonrolph]

Hi all,

I have made a prototype GeoGuessr style game for iNaturalist which you can access here: https://simonrolph.github.io/iNatGuessr/

It gives you some photos of some observations and then asks you to guess where they are from. I wanted to make a prototype to see if it was actually a viable game (it was also suggested here: https://forum.inaturalist.org/t/if-anyone-is-interested-in-making-a-game/34440). I personally have found it quite difficult to guess the location even vaguely correctly! But I want to get your opinions.

• Does the concept even work? Too difficult?
• What other clues might you want? Species names?
• Include more / less photos?

If you want to restrict the geographic range then you can add ?place_id=X to the url where X is the place_id. For example https://simonrolph.github.io/iNatGuessr/?place_id=6857 will filter only for the UK.

Also want to shout out BioGuessr: https://forum.inaturalist.org/t/biodiversity-education-quiz-game/24454

Please note this is a very early stage proof of concept and is likely to not always work and have limited features, just wanted to put this out there and see what you think. I wrote most of it with help from ChatGPT. It probably isn’t very mobile friendly yet.

[neontetraploid]

even from the first batch of photos I can already tell unmarked cultivated plants are going to make this extremely difficult lmao

[neontetraploid]

also this might be an issue

[jamfries]

It’s hard! Maybe if I was a better identifier it would be easier though lol

I think if I had a suggestion, it would be to try and get more of a variety of species in the images. The ones where they’re all images of butterflies or something are trickier than those with insects/mammals/plants

[gatorhawk]

A challenge for sure, but nonetheless awesome! Thanks!

[carnifex]

In the first three rounds, I guessed the correct country.
What I noticed is that in each round, several observations came from the same user.

I’d say it is very important that there is a diversity of users, and preferentially also different seasons.

How are the places chosen? Are these random numbers or are they restricted to a certain level (like county)?

[pisum]

@simonrolph

Does the concept even work?

yes, i like the bones of it, but…

@simonrolph

Too difficult?

@simonrolph

What other clues might you want? Species names?

right now, the clue that often gets me closest is actually the observer name(s), which is think is not how the game should be played. i think this would allow for better gameplay:

@pisum

i think the way i would do it, the game simply picks a place and then presents 5-10 of the most popular taxa from each of these categories:

• vertebrates (last 30 days)
• invertebrates (last 30 days)
• plants and other organisms (last 30 days)
• vertebrates (last year)
• invertebrates (last year)
• plants and other organisms (last year)

you could click on any of the taxa from any category, and that would present you a random observation of that taxon from the appropriate taxon + time category (last 30 days or last year).

[fmiudo]

I suppose the final version will have a time limit for the answer. It should be something quick too otherwise people will try to use the CV to pinpoint the whereabouts. Otherwise it is great! Good job.

[F.S]

This is super fun!
I got a pretty good score

The only issue I noticed is that sometimes the credits make it too easy to guess even if the observations themselves are hard to locate.
Username language is a huge cue, and one of the rounds was full of photos from the user “appalachianwonderer”. Needless to say the observations were from the Appalachian mountains.
Maybe credits could be obscured until a guess has been made?

[pttrsn]

I really like it and I agree with other’s suggestions of hiding the username credits until after the guess.

I’m not sure how much can be done with forcing different users for each photo, if the locations are going to truly be random. Sometimes there just aren’t that many observers in an area.

[gijsroaming]

I really enjoy the concept as well! Not too difficult if you’re ok with never getting a perfect score ^^.

Like some others have said being limited to 1 or 2 observers might limit the variety of types of species that you see. Perhaps taking a slightly larger range would be an option to resolve this.

I do think the names might be giving away too much. Whenever I was not sure about which country in Europe I was in I could usually figure out which it was just by the sound of the name of the observer.

[anon83178471]

I love GeoGuessr! I’m terrible at it! And was not very good at this either but it’s still a hoot!

[onyxrat]

I love the idea.
Might need to mix up the taxons represented more though. More than once many of the photos were of the same taxon from the same user.

[calibas]

I would make it so it picks at least one mammal, reptile/amphibian, bird, fungus, insect and plant observation.

It should exclude things that have global distribution and generally look like same (see tribe Astereae).

So a greater focus on things that are more recognizable to your average person, like mammals, and fewer things that could be anywhere, like white asters.

Edit: Also, have the images link to the iNat observation after you’ve made a guess.

[eh111]

Love it!
feature changes,
-show usernames after guess made
-link to iNat after guess
-A feature that randomizes the date of photos picked from a certain area instead of taking out a single chunk of time. (I got one that was all pictures of the same bug over and over again, also huge panels of moths since they usually get uploaded in bulk)0
-A feature that averages your scores across all of your rounds to get a overall score
-Maybe a way to limit the guessing to a particular country state or province (this would mess with scoring a bit)

• A way to limit to a certain kind of life (eg. Reptile, plant, fungi, quercus) for extra challenge (this might mess up scoring also)
  Great application, I really enjoy it!

[DanielAustin]

Cool idea, nice prototype.

One improvement needs to be making sure species aren’t repeated as I just got one where half of the clues were Canada geese!

[scharf]

Oh, this is tons of fun.

Some taxa are dead giveaways, for instance, hummingbirds (the Americas), storks (Eurasia), onychophorans (Caribbean).

[zygy]

Pretty fun! My high score is 211km :)

[zygy]

It needs an option to show only animals or plants. My score typically depends heavily on how many animals are shown!

[pisum]

one thought… i would be a little worried that iNatGuessr might tread a little too close to trademarks for iNaturalist and GeoGuessr. i don’t think anyone has trademarked Gerald the Muskrat though. so maybe calling it something like “Where’s Gerald?” might be nice. i can imagine a little narrative that goes something like, “Gerald looks around and sees… Where’s Gerald?” then if you guess within n km, maybe you get some sort of token – for example, within 50km, you get a muskrat; within 500km, you get a muskrat track; and within 5000 km, you get muskrat scat.

another thought… i’m looking at the API requests made by the page, and i think there are a few things that probably need to be changed. here are examples of what i think are the main requests:

• https://api.inaturalist.org/v1/observations?day=27&per_page=1&page=42&captive=false&geoprivacy=open&quality_grade=research&photos=true&geo=trueundefined
• https://api.inaturalist.org/v1/observations?lat=18.2716334181&lng=-65.8304664003&radius=10&not_id=179898341&per_page=7&captive=false&geoprivacy=open&quality_grade=research&photos=true&geo=trueundefined

notes:

1. there’s an extra undefined at the end of each of these requests, which effectively cancels the geo=true parameter.
2. if you’re going to use geoprivacy=open, you probably also need a taxon_geoprivacy=open parameter
3. you may also want to apply a spam=false parameter
4. at some point, you may also need to consider observation and photo licenses, too.