Last month we had a pretty massive blogging campaign, getting quite a few posts to the top of Hacker News in a month. The results of the most recent TIOBE index shows that this has made a sizable dent in the rankings:
We were 43rd before, and now we are at 35! In terms of the search percentages, that’s 0.2%, or 50% more than what we had before. This means that our writing about GPU computing, differential equations, and machine learning has a measurable impact on language visibility and adoption. I hope this is encouraging for those who have already started showing what Julia can do, and if you haven’t yet, give blogging a try! You can get started by posting links to here and by submitting your feed to JuliaBloggers.com:
The big outlet is then Hacker News, where a post can hit the front page if it grabs enough buzz. Feel free to let others know what to search for on Hacker News to find your thread by alerting people in the chatrooms:
I would like to contribute to that, but I don’t have a blog or want to start one…
If I make a big markdown file, can I somehow send it to juliabloggers.com?
I checked the page submit rss feed, but if I understood correctly that requires me to already have a blog?
Here are a few ways to get started. The easiest way is to make an account with something like Wordpress.com or Medium.com which are big blogging sites and you just make your articles in them. Simon’s GPU computing article is an example of this: Writing extendable and hardware agnostic GPU libraries | by Makie | techburst | Medium . If you want more control, you can also create a Wordpress blog on your own host (which is what I did for StochasticLifestyle.com), I recommend Bluehost.com’s cloud hosting because you can adjust your computing power to deal with surges (I had to do this when a few posts went to the top of Hacker News since that can get you >10,000 views in a few hours).
Another way to do it is through Github. The JuliaDiffEq blog is hosted through a simple Jekyll page:
You then just need a host if you want to setup this with a domain, which I use Bluehost. Then you just need to setup the domain forwarding, which is described here: Managing a custom domain for your GitHub Pages site - GitHub Docs . There’s a tab in CPanel to do this kind of stuff, and this is how JuliaPlots, JunoLab, etc. all use a special domain name yet are simply a Github repo.
While blogging is indeed a great way to spread the word, some minor things about JuliaBloggers.com are broken (eg some blogs have broken lines in the RSS feed, MathJax support is flaky because delimiters are not escaped properly, etc).
I tried contacting Randy Zwitch about them repeatedly, with concrete suggestions for solutions, but got no replies after a while. Perhaps as the language community grows we should rethink blogging infrastructure — this is a problem that many communities have robust solutions for, eg using an aggregator like Planet, which I am not going to link because the website seems to be down. Looking at feed aggregators, pluto seems to be maintained.
Hugo is another great static website generator, it is lightning fast, with Mmark support which I find very convenient for blogging about technical issues (seamless math, code integration). The Hugo docs has detailed instructions about setting up your blog, it will probably take a few hours. If that helps, you can find the source of my blog in this repo.
The TIOBE index measures the number of hits for +"<language> programming". For Julia, the <language> could be Julia, Julialang or julia-lang. This is described in the TIOBE definition.
It is interesting that the query "JuliaBloggers" + "Julia language" gives 865 hits in Google, while "JuliaBloggers" + "Julia programming" gives 527 hits. In Google, "Julia programming" has 177,000 hits and "Julia language" has 46,000. Interestingly, "Julia programming language" has 104,000 hits.
Is our use of Julia language instead of Julia programming lowering our TIOBE score?
They might be amenable to including things like that, for all languages. I think it would be useful for all of us in the Julia community to sent e-mails to the people at TIOBE, requesting these changes:
From their site:
This is the top 5 of most requested changes and bugs. If you have any suggestions how to improve the index don’t hesitate to send an e-mail to tpci@tiobe.com.
Apart from “ programming”, also other queries such as “programming with ”, “ development” and “ coding” should be tried out.
We also need to let them know that esp. for languages like Julia, Go, Rust, Mumps (or Lua in Portuguese) where it’s a common word (unlike C#, Erlang, Perl, C++) that people often differentiate by having “ language”, and that really should be taken into consideration.
As I indicated on the other thread, I don’t think obsessively following these things with bated breath is good for anyone’s sanity. The fact that we see such wild fluctuations seems indicative of huge statistical uncertainties.
Fear not, Julia will stand on its own intrinsic merits. It’ll take a while, but things seem on track to me.
I’m not at all obsessed by it, I think it’s funny. The TIOBE index has a ton of problems with it’s ranking mechanisms - besides being a popularity contest, I’ve seen that it’s bad at miscounting languages whose names are hard to distinguish, and I don’t think it deals well with things not in English (since it looks for words like “language” and “programming” to distinguish).
IEEE has an index that I think is much better for gaining insight about languages.
Was this not something I fixed? I’m not a MathJax person, and I am reading everyone’s RSS feed using a stock WordPress plugin, so it’s not something I’m completely attuned to. The main limitation is someone with PHP experience (or, moving to a different blogging software)
They really need to use some modern NL processing on this! There are so many good ways finding out if a text contains a certain concept rather than this. They don’t really show themselves to be very good hackers, IMO!