Hacker News Reader

Comment by biot | Hacker News Reader

If you were made CEO of SourceForge, what would you do to make it relevant again? Assume shuttering the site isn't an option and you have to do your damnedest to make it work. What would it take? Beyond just pure, clean code and projects are there any services which sites like GitHub don't offer that might entice developers back, regardless of cost?

The only things I can come up with are services like a free server farm for built-in continuous integration and deployment. Maybe strike deals with Heroku, AWS, Azure, Google, Digital Ocean, etc. to make one-click setup and deployments from the project a reality. If they can demonstrate trustworthiness (some sort of trusted third-party auditing?) and rapidly execute on delivering massive value, they just might be able to regain some of what they lost.

View on HN

Well, I'll assume shuttering the site isn't an option. Be careful also because apparently using the word "the" before the word "site" is taboo to some commenters here. We're going to do the obvious stuff first: get rid of bundled software (done), get rid of deceptive advertising, improve features (like rolling out https, done), and then we will look at ideas like you mentioned. I appreciate the feedback.

Yes, you would gain a massive chunk of traffic if you can get your site spam and malware free. There are still some projects like mp3gain, etc that are useful.

Right now uBlock won't even let me visit Sourceforge without adding it to my exceptions list (probably for a good reason).

Yes the malware is gone. Now what's left is somewhat deceptive ads in some cases (ads with download button images in them). These are served programmatically from ad networks but we're building a system where they can be reported and acted on right away. Hopefully we'll get uBlock to come around.

All of your competitors have zero ads. Are you exploring potential revenue models other than advertisement?

Yes but nothing in the way of unwanted bundled software or other models that would upset developers or users.

One feature of the old Sourceforge that I found helpful was the "compile farm" - not so much for the "compile" part (cross-compilers are nice), but to test the newly compiled code.

The reason I chose Sourceforge for all of my projects in the distant past was because of their dedication to Open Source and Free software, not any particular technical merits (though, technically, it was reasonably functional for the time, but a decade of no significant improvement is hard to miss).

Github is a great product, but they are not as open as SourceForge once was (the source code to run your own SourceForge was available).

Also, SF still has several tens of thousands of projects, including a few very large ones. Merely stopping the bleeding would be sufficient to remain a large OSS project hosting site.

Thanks for the support. Stopping the bleeding is what we're doing now and then we hope to improve SourceForge significantly.

The zeroth step would be to have a vision for sourceforge.

The first step would have to be slim down the site to make it actually fun to use. It tries to be everything for everyone. It has so many options and features, but most of them with such a bad UI that it takes ages to accomplish anything.

I remember that I once found a bug report on a project I used, and even after logging in to sourceforge, I couldn't find a way to submit a patch or add a comment to that bug report. That's simply a no go. Make it easy to contribute, even if that means throwing out half of the subsystems.

And yes, something CI/CD-like might be a good vision. Adding it to the existing cruft would not create a good experience though.

I wrote about this before once [1], but the key thing SourceForge needs to realize is that the users content is the single most important asset they have, and they need to present that on the page with more importance than any other UI element. It's something GitHub got right from day 1, but SF never managed.

[1] http://www.codersnotes.com/notes/the-winter-of-our-disconten...

Even if shuttering the site were an option... it would be a bad one. The site has millions of daily visitors and is within the top 300 worldwide. There's ad revenue in that alone if you wanted to go that route.

I wrote that mostly tongue in cheek, given that I suspect more than a few people here on HN would, if made CEO, shut the whole thing down as a public service.

Support for all the package managers for the most frequently languages out there. I still stumble over libraries hosted on sf.net a lot.

Good idea

If I was in it for the really long haul I would try to strike an integration deal with gitlab, and position myself as part of the ecosystem with gitlab for the source, and sourceforge for the binaries / large assets, and (try to) make money with commercial developers purchasing more private space for their assets (say free tier is x gb, you can pay for much more)

Not sure it would work, but I think having a counterpart to gitlab for the binaries/assets and for the non-technical users in general would be nice

Reply on news.ycombinator.com