It could be a curse or popularity? Some projects are so popular that you can place them nearly anywhere, and they would still find a lot of densely connected group of neighbors
> A lot of country labels were generated with help of ChatGPT. If you find something wrong, you can right click it, edit, and send a pull request - I'd be grateful.
What else did you use to calculate similarities? When I tried something similar with HN co-commenters, I just got one big blob with dang and other prolific commenters in the middle. That was not very interesting or insightful.
How long did it take to calculate the Leiden clustering and the force-layout? Do you think it would be somehow possible to compute a force-layout of the whole graph?
This is amazing, the GitHub ham radio and hardware scenes are way bigger than I previously thought. Shout-out to OpenRTX and M17 along with all the awesome SDR applications!
This is amazing! I love this effect of showing me that my little project that I was working on fairly solitarily is actually part of a community of other people doing similar work that I can reach out to/collaborate with etc!
What does it mean if, I click on a repo, and it shows 5-6 links to specific projects? Does that just mean the jaccard similarity index was below a threshold?
Yes! I picked only the highest scores to form an edge in relationships graph. Typically a sigma (std deviation) or two away from the mean. So if there is a direct link between your project and others - the similarities are abnormally high.
Note that I'm not rendering direct links outside of the country yet, there might be more there. Will probably add a "focused" view to see those better
I've been watching changelog nightly emails for a while - they summarize most starred daily repositories, and growth of AI there (subjectively) seems to be even higher than frontend tech
To me this is proof that doing a good visualization is really an art. I saw this on hn for a couple hours, but really didn't think much of it. As soon as I clicked through the funny names, and "continents" brought a smile to my face.
I guess my little project is too small to make it, but now I too aspire to join the great nation of Golandia.
I bet that several of these regions have a common image in their readme (the python logo, the nix logo, etc). Imagine little flags popping out of each region...
haha, I love the flags idea :). I wish I had the prowess to implement it in a way that is visually appealing and not obscuring the map.
I would also love to have a giant octocat hugging the archipelago, with some radial gradient emitting inside of it. Alas my design-gl foo is not there yet
1. this is so cool. i'd love to see more name for the regions!
2. nit: one repo dear to my heart dbt-labs/dbt-core was previously dbt-labs/dbt as well as fishtownanalytics/dbt. they show up as unique nodes on your map. what most interesting to me is that each name of this repo links to a different set of related repos?
The connections are inferred from stargazers. If they lead to different set of related projects it might be a sign that different group of people gave stars different things at that time.
Github of course has much more dimensions than a flat 2d surface can show
This reveals an interesting dichotomy in the project I work on, Hail. I think of hail as a serverless workflow, relational, and linear algebra system most similar to Dask, BigQuery, Snowflake, Spark, etc. but this map is constructed from the perspective of the user so Hail lands squarely in the world of bioinformatics.
Also neat to see how bioinformatics is such a splitbrained community. They land next to R but are filled with Python projects.
If you work on a dynamic version that allows users to understand changes in the open-source topography over time and detect/predict new clusters, this could be a very powerful tool for investment intelligence.
I love these things - to see everything sort of classified, and all at once, helps me find things I would never think to look for. Forgive me if this has been posted somewhere else, but if you want to see a reddit map, here it is (maybe you found it already in their github, but I just have it as a bookmark): https://anvaka.github.io/map-of-reddit/?x=18083.096950551575...
Yeah I just wanna paste this and found your answer :P, now we have another similar tool about Github, much more useful for many people, kudos to the author!
Vue is a pretty big deal in the PHP/Laravel community. Evan the creator even gave a talk at Laracon a few months in, showing the hockey stick growth after it started to get traction in the community.
True. Though I also expected to "Reactistan" somewhere more explicitly, I wonder if it is scattered in Fronterra continent?
I am also puzzled myself why the algorithm has separated Fronterra from another large island to the west, seemingly related to node and some other JS libraries. Decided to keep it all as is in case I'm missing some reason
My feeling is that the React community is at least double to triple the size of the Vue one, so yes, I feel React should definitely be it's own island.
84 comments
[ 2.8 ms ] story [ 145 ms ] threadNote that some country are missing their names because I didn't have enough knowledge to assign a name to them.
Here is my naming process if anyone wants to help with the rest: https://github.com/anvaka/map-of-github#country-names
But ClickHouse itself somehow appears in Kubernation, and ClickBench - in Datapolis. Nice names btw.
It could be a curse or popularity? Some projects are so popular that you can place them nearly anywhere, and they would still find a lot of densely connected group of neighbors
> A lot of country labels were generated with help of ChatGPT. If you find something wrong, you can right click it, edit, and send a pull request - I'd be grateful.
But how do I zoom in/out?
Seems like a really overreaching power. Wait, never mind, that makes sense.
There are a lot of interests that I didn't know exist. For example https://github.com/cat-milk/Anime-Girls-Holding-Programming-... - someone collects anime girls holding programming books.
https://github.com/tylertreat/Comcast - and here is someone who is amazing at coming up with funny project names =)
How long did it take to calculate the Leiden clustering and the force-layout? Do you think it would be somehow possible to compute a force-layout of the whole graph?
I know this because I moved visionmedia/debug to debug-js/debug in November of 2021.
https://cloud.google.com/blog/topics/public-datasets/github-...
It explains the full pipeline - how to download, collect, and analyze this sort of data.
What does it mean if, I click on a repo, and it shows 5-6 links to specific projects? Does that just mean the jaccard similarity index was below a threshold?
Note that I'm not rendering direct links outside of the country yet, there might be more there. Will probably add a "focused" view to see those better
Some say it's a cancer others say it's inevitable and here to stay. Probably both are right.
Also, why is Swiftoria so big?
I guess my little project is too small to make it, but now I too aspire to join the great nation of Golandia.
I was waiting for about 10 minutes wondering how big these indexes were...
Though I noticed a couple of odd groupings - like that MicroPython was clustered in Arduinoria rather than the adjacent MicroPythonia... :P
I bet that several of these regions have a common image in their readme (the python logo, the nix logo, etc). Imagine little flags popping out of each region...
I would also love to have a giant octocat hugging the archipelago, with some radial gradient emitting inside of it. Alas my design-gl foo is not there yet
The connections are inferred from stargazers. If they lead to different set of related projects it might be a sign that different group of people gave stars different things at that time.
Github of course has much more dimensions than a flat 2d surface can show
Also neat to see how bioinformatics is such a splitbrained community. They land next to R but are filled with Python projects.
If you work on a dynamic version that allows users to understand changes in the open-source topography over time and detect/predict new clusters, this could be a very powerful tool for investment intelligence.
I am also puzzled myself why the algorithm has separated Fronterra from another large island to the west, seemingly related to node and some other JS libraries. Decided to keep it all as is in case I'm missing some reason