[ajdlinux]

Site's not loading very well for me.

I definitely like the idea of preserving the source code of abandoned projects though. Ultimately, I'd like some equivalent of the "legal deposit" principle that applies to books and other printed works - some kind of scheme where source code, not just object code, for proprietary products eventually makes its way into the public domain. I'm not sure how this scheme would work, but I'd like it to happen.

[IgorPartola]

I worked for a company where one of the clients required that all the source code be periodically archived and shipped to them so that if we went out of business they could potentially run it themselves. Practically, they wouldn't be able to because of the sheer complexity of all the different services. It was a minor pain to do this for them but the contract was worth it.

Having said that, I am a big fan of open sourcing projects of they fail. Why not? What do you have to lose?

[david-given]

Well, judging from my experience...

- failed startups don't just stop, they go into administration and get auctioned off, which means someone else owns the IP.

- proprietary source usually contains licensed copies of other people's source, and sorting out who owns what and what's safe to release requires a surprising amount of time and money.

My previous place had a tonne of interesting IP, all of which is now owned by a VC company who will never, ever do anything with it. I'd have liked to open source it if we could have, but it was a total nonstarter. shurg

[bigtunacan]

I've never worked for a company where we shipped the code directly to the client; that seems a bit risky for the company producing the code.

I did work for a startup though where all of the code was backed up and sent offsite to an escrow service. It was part of our standard contract that if the company ceased operations all clients would have received copies of the code. This was a pretty big deal to quite a few clients as one of the biggest risks for a larger client investing in using software from a small company is lost sunk costs if the startup fails. If the client has this type of agreement in place it is much less risky. If the startup fails then a heavily entrenched client would request to receive the code and more often than not hire one or more of the now unemployed software developers either as employees or consultants to either maintain the software or train someone else to.

@frazras actually from this context since I believe it is a fairly common problem, the closed casket paid service makes a lot of sense. What would be ideal is if it could essentially be turned into a GitHub/BitBucket (private) style service that is dead simple to use, but also acted as an escrow service. You would need a means of tracking clients, but not allowing them access unless terms of contract were met that allows access. Then it would also need to lock down the codebase in some manner; don't allow repo deletes, do backup snapshots, something like this so you can guarantee the startup doesn't breach contract.

[IgorPartola]

I think you are actually right, it was done via Iron Mountain. I wasn't directly involved with the process, only was aware of it.

[nirvdrum]

I wrote about why I opted not to open source my failed startup (and there was some lively HN discussion around it):

https://news.ycombinator.com/item?id=8641867

Eight months later, I still think it was absolutely the right move. Don't get me wrong, I wish the project could have lived on. But I sleep much better at night knowing I didn't open myself up to any additional liability or feeling inadequate for not having the time to spend helping others do something worthwhile with the code.

[ajdlinux]

I was involved in a project which had some potential to become a startup, but for various reasons we decided against going into business - we ended up GPLing the proof-of-concept application we'd been working on and put it up on GitHub: https://github.com/tsujamin/digi-approval

[x5n1]

Because most of the time it's useless to open source projects that don't compile and run. Though perhaps I am sure there might be someone who might care enough and move in and make it run. I think something like that happened with Tesseract, Google stepped in to make something than was not running, viable.

[richardwhiuk]

Could be part of an escrow contract - e.g. if the company goes out of business, after x years, the escrow agent will release the code into the public domain.

[A_Beer_Clinked]

I've been through this a few times as both the vendor and the client. There are normally a bunch of conditions that can trigger a release of the the source code to the client. These _could_ include business failure but also non-performance or acquisition.

I've never come across one where escrow agent will release source to the Public Domain but I guess it could happen.

Companies like Iron Mountain have escrow services for clients of SaaS providers. I think in these cases data escrow is significantly more important than source code escrow.

[at-fates-hands]

I was thinking about something similar.

If a failed startup releases their source code and another team picks it up and solves whatever problem the original team had - time, resources, bugs, etc. and is successful, then there would be some kind of kickbaack (albeit small) to the original owners or developers.

You could think of it as a code consignment shop.