This brings up other issues that are generally the main reasons people build web apps in the first place:

1) Updating code is a pain. You can't just push a patch, you have to rely on others to download the patches and install them

2) You now have to support multiple versions. Because of above, you'll be getting bug reports for old versions

3) Different hosting environments. Not only wil you end up with support for environments you've never heard of, but you'll have to contend with bug reports that are actually to do with misconfigured servers.

4) Unless you are targeting enterprise (in which case there IS a business case for self hosting) the number of organizations with the know how and resources is probably far lower than the rest of your user base - and they will probably take up most of your support time.

5) You lose the subscription model. Certainly some apps can charge people on going fees for support etc, but subscriptions are such an easy way to make money in software.

Your point is valid, but it kind of goes against why people chose to build web apps!

A simple solution would be to provide a virtual machine image with the web app and environment preinstalled and only provide support for that. If people want to host in their own environment then they can debug issues themselves. If they want to report a bug then they must test it in the VM instance first. This removes the majority of the disadvantages.

Not to mention the fact that there's a whole new level of security to try to prevent sharing of code and folks using it past licensure dates that you don't need to worry about with a subscription/SaaS model.

The application could "phone home" occasionally to check up on the status of the license. Should the business close, an override code could be provided (and fulfilment of this could be a contractual obligation).

Applications written in interpreted languages can usually be secured with tools built for the purpose (e.g. ZendGuard for PHP). Compiled languages do not have this problem in the first place.

How do you enforce contractual obligations against a business that no longer exists?

Have a contract directly with the owners, I guess.

I'm going to ignore the issue of the extra costs this imposes on the startup, and ask: what proportion of the target market are actually interested in self-hosting?

The people you (or anyone else) interact with online are not representative of the broader public.

Sorry - I've had two cups of coffee and I'm still not understanding what you consider "the biggest need web startups fail to address". You said:

"The single most important factor which makes me gloss over new web startups and the services that they provide is that there is no option for self-hosting."

Self-hosting what exactly? What are we even talking about?

Wait - upon my fourth reading, I'm guessing you are referring to "web startups" who offer an API or other SaaS offering. Is that what you are talking about? I'd rather ask than spend 20 minutes writing a response to that and then for you to later so, "No, that's not what I meant."

Yes, I think you grasped his post. Software you use in a relative isolation (yourself only or with a well-defined group) could be installed and maintained on your server. He claims he'd look into web startup offers more if they provided this alternative.