Nice post! I love hacking around with PostGIS. While I'm not normally a huge ORM advocate I do like the GeoDjango API for this sort of thing [0].
There hasn't been enough experimentation with geospatial social applications. (There was a lot back in the early 2010s but it petered out.) I'm currently working on an experimental chat app that creates Slack-like channels associated with a topic and a location. I'm looking for collaborators on this project (primary need is for designers and community-builders, but also programmers) so if it appeals to you please get in touch (HN-facing email in profile).
Very interesting, and good intro to Postgres R-trees! I kind of miss YikYak, it was pretty fun to see all the posts around UCLA. If it focused a bit more on promoting good content (and not just mindless noise, which eventually turns into edgy racist garbage), it might still be around.
(As a side note, I've always wanted to give MongoDB geospatial[1] queries a try. Wonder how they compare to Postgres' offering.)
Hey - I'm the author (I self-posted this, I hope that isn't against the posting etiquette here).
I wrote this post after a bit of a failed startup I tried making at the start of lockdown. It was called Ottr and it essentially aped YikYak exactly. I made a react web app, an API with web sockets (it had real time posting and comments) and even an iOS app.
I burnt out on it and gave up because it seems so hard to get that initial traction.
I've been trying to get the project posted publicly on GitHub for a while now but I wrote maybe 10k+ lines with literally zero tests because I was just trying to get it to market. The code isn't bad but i'd feel bad posting it without tests so that's what i'm working through now.
So I had a bit of a theory (it's stupid I know - you don't need to tell me) that meant that YikYak was to Ottr as Vine was to TikTok - as in;
The idea behind YikYak was so good it was ripe for someone to come along and play of the nostalgia that I (and many of the people I know) for it. YikYak came along when I was in first year of university and it absolutely blew me away with how inventive the idea was. I had so much fun with it, and people were creative and funny on the app.
There is something about it that tickled that tribal part of your brain because it felt like you were part of a big community (but not too big that you feel alone). Seeing a post about your actual friend on the feed from someone anonymous was so engaging.
Lockdown seemed like such an opportune time to release a hyperlocal social network, but I spent so much time tweaking it and trying to make it perfect, but I didn't know how to make people actually use it (I still don't know).
I had Ottr in a working state, but i've made peace that it is solely a learning experience for me, and a fun one at that! (until the burnout of course)
Don't be afraid to put unfinished code or code without tests up on GitHub -- just mention in your README that it's not production-ready yet or something. It still has educational value for the community. Maybe someone will even help you write tests! Btw, thanks for sharing -- I had never heard of the R-Tree data structure before.
I really should just do this. So there are a few parts (API, App, Web, DB) to it - I wonder what is the best way of representing that? Just have the same readme with links out to the other projects maybe?
Don't know if they're still up but I remember there were a bunch of yikyak clones around the time it stopped being popular. Just weren't enough people in any of them, so I stopped them. Network effects and all.
Should have created an open YikYak protocol and merged them all into one herd.
YikYak was for those a little younger than me but I found it really funny to watch and vote and it had the feel of the old internet where no one knew you were a dog.
Nice! I hadn't heard about R-Trees before. I guess you can potentially make do with a quadtree[1] but it seems R-Trees are better suited for geodetic data.
It's so weird how yikyak blew up then rapidly failed. I used it pretty heavily in undergrad and popularity didn't seem to wane during that school year when it got big. It just died over the summer since you were geolocked to your hometown, then it inexplicably never picked up again the next school year, just dead on arrival. Had they let people still use their collegiate yikak while off campus for the summer, yik yak might have been a juggernaut by today. I still don't think the niche has been captured: an anonymous, hyperlocal twitter/reddit hodgepodge, fueled solely by original and genuine campus memes and therefore immune to corporate ad peddling like global networks. Maybe that's why it failed.
For a while and probably until it died, they actually did allow you to use your college yikyak when you were away. I think you could use two locations, your current location and your "home" location. It died overnight because they forced everyone to have usernames one day. Probably because they wanted to monetize.
There was a lot of social/cultural issues with YikYak. My college explicitly forbid using it, and a lot of other small colleges had similar policies, largely because of abusive posts happening on the platform.
I know they ended up working with high schools to limit access in certain areas. But I'd be really suprised if a college could legally forbid using the service.
Their opaque moderation tactics also killed it. There were the transparent obscenity warnings, but then there were also shadow downvotes for certain terms that even closely resembled a human name, which was rife with false positives. You could talk about a street with a human name and 10 minutes later your comment was voted out of existence.
Yep - YikYak did let you do this after a while, but if you check the subreddit for YikYak they rolled this change out along with a feature that meant you had to give or an email or phone number IIRC. Huge backlash and people stopped using it.
I tried to solve this in Ottr but letting a "Explore". They could open a map, drop a pin and that would be their new location. If they posted while "Exploring" their post would be marked as such.
I interviewed at YikYak shortly before they released the feature that killed them. I actually got to talk to the CEO at some point. They discussed adding something personally identifying to fight bulling, which was a huge problem with YY at the time. I also asked how they planned to monetize, but honestly the response I got didn't seem very fleshed out or just not thought through very well.
Sucks because they actually had some very impressive engineering going on at the time, and I really wanted to absorb some knowledge from those guys, but unfortunately it didn't work out. YikYak shutdown less than a year after my interview.
Interesting post! I also recreated YikYak as a side project and had to solve a lot of similar problems (It's called Clacku on the iOS and Android app stores if anyone cares). Managed to get a decent chunk of users to actually use it.
24 comments
[ 3.2 ms ] story [ 65.7 ms ] threadThere hasn't been enough experimentation with geospatial social applications. (There was a lot back in the early 2010s but it petered out.) I'm currently working on an experimental chat app that creates Slack-like channels associated with a topic and a location. I'm looking for collaborators on this project (primary need is for designers and community-builders, but also programmers) so if it appeals to you please get in touch (HN-facing email in profile).
[0] https://docs.djangoproject.com/en/3.0/ref/contrib/gis/
(As a side note, I've always wanted to give MongoDB geospatial[1] queries a try. Wonder how they compare to Postgres' offering.)
[1] https://docs.mongodb.com/manual/geospatial-queries/
I wrote this post after a bit of a failed startup I tried making at the start of lockdown. It was called Ottr and it essentially aped YikYak exactly. I made a react web app, an API with web sockets (it had real time posting and comments) and even an iOS app.
I burnt out on it and gave up because it seems so hard to get that initial traction.
I've been trying to get the project posted publicly on GitHub for a while now but I wrote maybe 10k+ lines with literally zero tests because I was just trying to get it to market. The code isn't bad but i'd feel bad posting it without tests so that's what i'm working through now.
The idea behind YikYak was so good it was ripe for someone to come along and play of the nostalgia that I (and many of the people I know) for it. YikYak came along when I was in first year of university and it absolutely blew me away with how inventive the idea was. I had so much fun with it, and people were creative and funny on the app.
There is something about it that tickled that tribal part of your brain because it felt like you were part of a big community (but not too big that you feel alone). Seeing a post about your actual friend on the feed from someone anonymous was so engaging.
Lockdown seemed like such an opportune time to release a hyperlocal social network, but I spent so much time tweaking it and trying to make it perfect, but I didn't know how to make people actually use it (I still don't know).
I had Ottr in a working state, but i've made peace that it is solely a learning experience for me, and a fun one at that! (until the burnout of course)
YikYak was for those a little younger than me but I found it really funny to watch and vote and it had the feel of the old internet where no one knew you were a dog.
[1] https://en.wikipedia.org/wiki/Quadtree
I tried to solve this in Ottr but letting a "Explore". They could open a map, drop a pin and that would be their new location. If they posted while "Exploring" their post would be marked as such.
Sucks because they actually had some very impressive engineering going on at the time, and I really wanted to absorb some knowledge from those guys, but unfortunately it didn't work out. YikYak shutdown less than a year after my interview.