The technical infrastructure of the major open source projects would make a small startup cry. It's certainly not out of the reach of Open Source. Just knowing the infrastructure required to keep the KDE project ticking, it's far beyond what Xobni would have to deal with.
If we're thinking in terms of "Open Source applied to services", well, Wikipedia. That's a pretty big website thinger there. I'm not sure I'd want to stare down the sysadmins of WikiMedia an inform them, "No guys, no, see, you're not really hosting..."
I think he meant to say "Ask yourself: what open source software exists that solves a big problem in a large market?" ... then host it and provide it as a service ...
Yes, I agree with your addendum. This is what we have done with Akvo.org. Nearly 3 billion people do not have clean water to drink or adequate sanitation. That is what I call a large market and a big problem. We have taken the open source approach to try to address this from an internet service point of view. Even though we haven't released any code yet, we have several big players investigating using our platform and software for boosting their efforts in the development aid space. Maybe we are doing something right.
"... "Ask yourself: what open source software exists that solves a big problem in a large market?" ... then host it and provide it as a service ..."
I'd agree with that one.
Free software allows you to take an individual copy then run it yourself or even run a company. It's not as if Free or Open source software has to serve the market any more than that.
nah, you could sell that. vnc/rdesktop in an applet, set up to let you test your site in any browser/os combination. Like browsershots.org, but interactive.
I see at least two people didn't understand my comment. The point of the post was that you can make money by finding some useful open source software and making it into a service. But Linux is already available as a service; that's why I said try again.
I think a more interesting question is: can we work around this?
I really don't know what the state of P2P or decentralized systems are, so tell me if this sounds crazy or not. Maybe someone could develop a secure, trusted, reliable system for distributed execution of sandboxed code which relies on the BitTorrent sort of sharing is caring strategy. That'd be pretty sweet :-)
Building the system would be easier than solving the problem of knowing who to put in jail when it gets infested with the usual junk that tends to end up in anonymous systems.
But it'd be useful, so please go ahead and solve it :)
Or, just maybe, a system such as this would prove once and for all that it just isn't worth trying to control, censor, and otherwise police the internet. A hacker can dream, can't he?
Well, -someone- is going to jail when The Authorities find illegal content (whatever that means, the concept seems to be expanding quickly) on your computer. Would you prefer it to be you, or the one responsible for it? Unless you can solve that problem, I'm pretty sure you're going to have trouble finding people to volunteer.
As for policing the internet, I think that personal responsibility for your actions needs to apply even when you act using a computer. The consequences of your actions should to be reasonable though, and not like what we're seeing now with file sharing lawsuits.
That kind of defense hasn't worked very well in filesharing cases (or has it? at least I'm under the impression that it hasn't), and is even less likely to work if the offense is more severe. Like when the infamous "terrorist networks" put things there, or the kiddie porn people. At the very least you'll get hit for aiding the baddies.
In freenet you generally can not see what is stored on your node (i.e. PC) because it is encrypted. To make this work, a freenet 'URL' more or less contains the key to decrypt the content it references.
Sure, but you're still helping someone do something by taking part in the network. If that someone is doing something illegal, you're well on your way to getting in trouble. Especially if most of the activity in the network is illegal, then it'll be hard to argue that you didn't know what could happen. It's pretty much what killed Napster, although in that case it was the company that ran the network that got shut down.
I suppose it'll have to be a serious problem for someone to make the effort of finding and decoding the data, but I don't doubt that sooner or later someone will decide that it's worth it.
If there's going to be processing of the data, and not just storage, the local node will need to have access to the encryption keys. That should make the work of the investigator easier. With enough cunning trickery it'll probably be hard to find them, but with access to the hardware I have no doubt that it's doable.
Hm. So the work needs to be done in a virtual machine that is given access to a local file storage and the freeserve (let's call it that, sounds nice) network. The vm needs to query the network for tasks to be performed, download the data for the task, process it, and upload the result.. Should be a fun project to design it.
Fairly simple. Except for all the technical and legal problems :)
You should have a look at the freenet documentation. They thought about all those problems already. (And your guesses about their solutions are partly right, partly wrong. For example your node does not have to know the keys and no VM is necessary.)
If you're going to do any processing on the data you're going to need to decrypt it, and then you'll need the keys. And unless you -really- trust random people on the net, you'll probably want to run that processing in some sort of sandboxed environment. I went one step further and put it in a VM, since that should be easier to close down.
Just running a Freenet node doesn't it of course, but I was thinking of the 'peer-to-peer servers in the cloud' thing.
> If you're going to do any processing on the data you're going to need to decrypt it [...]
You underestimate modern cryptography. Or to give a simplified model: Imagine you just had blobs of encrypted data that you served and the requester would do all the processing.
"... Ask yourself: what open source software exists that solves a big problem in a large market? ..."
Statistical software (like R), perhaps? Not sure how much it's used by businesses though, especially with alternatives like SAS, or data mining built into Oracle & SQL Server. Still, hosted data analysis sounds cool (with potential complexities as well).
The guys behind SciPy are doing this sort of thing (not R, specifically, but scientific computing...they were doing fluid dynamics and geological data analysis). It's a pretty specialized market, but one where huge sums of money are spent. They had revenues in the millions when I left, and they only had two big customers (and a few customers who were putting out feelers for other options besides their current providers). Anyway, they weren't actually making any money at hosted scientific apps yet, but they were building their data center out and getting the contracts in place, so I expect it's a reasonable portion of their business today.
But, if you or your co-founder isn't a PhD in the field you choose to tackle, you'll probably have a hard time getting a foothold.
"We’d probably all be using an IM client built around Jabber."
Depending on your circle, some do. gChat is runs off the Jabber protocol. On the Mac, Adium is one of the best (I don't know about most popular) multi-system clients out there. I'm also partial to Pidgin Linux.
I don't personally need open source blogging or news aggregator software. And that is why I use blogger, google reader, etc.
But I do need Django, I do need libraries like ipaddr.py from Google, I do need Linux kernel, I do need Webkit. One by one the majority of those libraries don't solve big problems. But combined they do.
And as for developers, they can stop worrying that somebody will take their open source code and clone their major project and start releasing chunks of code that is actually interesting. Let's be clear, most of the code is pretty much the same everywhere and is not interesting.
Actually I think that GPL almost goes as far to encourage this hosting model. If you want to incorporate non open source code or libraries then you actually CAN'T release that code to the world. The only way to offer people that functionality is to host it.. and nobody else can offer a non-hosted solution. I can see a lot of startups following this model with GPL in particular.
"There’s no doubt we’d have better software today if open source projects could get servers and operational resources for free"
This is what heroku.com and seasidehosting.st offer. It does seem an opportunity for super low cost of entry apps. I don't know about how either scale if you get popular; presumably there comes a point where you have to rethink the infrastructure.
Interesting idea, but saying open source has this inherent limitation is overreaching a bit. A much more reasonable claim would be that open source "can't do" service-based applications (Which I'm still not sure I agree with).
"what open source software exists that solves a big problem in a large market?"
Linux, Apache, mySQL, etc.
It is an inherent limitation. Time is a cost that can be donated because there is some non-material payoff to the person that's donating the time. There's very little non-material payoff for donating servers, bandwidth, etc.
I actually like this. It's the most obvious way to capitalize on open source.
I can't help but wonder if it's worth it. The two main problems I see are 1) Something necessitating the AGPL & 2) Apps that don't need to be online (eg rescuetime) & the privacy concerns associated with them (eg bush & his telecoms).
I'm not sure if you're aiming to agree or disagree, but...yes totally :)
There are many other examples of the "open source business model," such as Dries Buytaert who runs Drupal and Acquia.
The point is that contributors to open source projects are willing to write code for free but not willing to wear the beepers that go off when the servers go down.
The point is that contributors to open source projects are willing to write code for free but not willing to wear the beepers that go off when the servers go down.
But could you get 100 people to wear beepers if you told them there was only a 1% chance it'd go off?
(It's a rhetorical question. I've seen what ten part-time sysadmins can do; I don't want to imagine 100.)
From the workflow I really like PipelineDeals. Unfortunately it's a bit buggy (especially the API). I just started the trial with them last week and if the stuff gets resolved by their customer service I'm pretty sure that's what we'll end up using. I'm planning a blog post which I'll submit here about the various CRM systems that we've tried out (notably SugarCRM, HighRise, Salesforce and PipelineDeals).
If we're thinking in terms of "Open Source applied to services", well, Wikipedia. That's a pretty big website thinger there. I'm not sure I'd want to stare down the sysadmins of WikiMedia an inform them, "No guys, no, see, you're not really hosting..."