29 September 2008

Watch Out! It's a Trap....

Go, RMS, go:

"One reason you should not use web applications to do your computing is that you lose control," he said. "It's just as bad as using a proprietary program. Do your own computing on your own computer with your copy of a freedom-respecting program. If you use a proprietary program or somebody else's web server, you're defenceless. You're putty in the hands of whoever developed that software."

Mind you, I think it would be better if RMS came up with ways of taming clouds rather than just excoriating them (assuming you can excoriate a cloud, which seems unlikely.)

14 comments:

  1. Anonymous5:55 pm

    I think it would be better if RMS came up with ways of taming clouds rather than just excoriating them

    I think he alludes to it here: If you use a proprietary program or somebody else's web server, you're defenceless

    To "tame the clouds", one must maintain one's own cloud. Software freedom requires that you are granted the necessary permissions to exercise absolute control over the operation of that software. If some other computer (like a web-server) does some of the processing, you need absolute control over that software too. I don't see any other way.

    It isn't really much different to deal with than stand-alone software that runs only on the PC. If you currently use proprietary stand-alone applications and you want to "go free", you must find or write your own replacement software. Same with web-based applications, the only difference is that you need to build or rent your own web-server to run the free server-side code on.

    ReplyDelete
  2. That's a bit defeatist, isn't it?

    Look here's Moody's patented FreeCloud system:

    you have a distributed set of computers running free software, that seamlessly run your apps for you, but these are run invisibly - nobody knows whose apps are running - and redundantly - several instances are out there, so you can never be held to ransom.

    There, why can't RMS work on something like that?

    ReplyDelete
  3. Given RMS's role these days as a law-hacker, isn't the GPL3 and Affero license, his attempt to tame clouds?

    ReplyDelete
  4. Yes, but it's always struck me as more of an adumbration of an idea than a solution: for example, it doesn't solves issues of data access and portability. I'm sure the collected wit and wisdom of the free software world can do better....

    ReplyDelete
  5. Anonymous8:51 pm

    That's a bit defeatist, isn't it?

    I dunno, how's it more defeatist than stand-alone free software?

    Look here's Moody's patented FreeCloud system:
    you have a distributed set of computers running free software, that seamlessly run your apps for you, but these are run invisibly - nobody knows whose apps are running - and redundantly - several instances are out there, so you can never be held to ransom.


    Sounds like you just described Google Apps (minus the "freeness" of the server-side software). I don't think I'm following?

    Even if the FSF ran the "distributed set of computers running free software", I still wouldn't have freedom over my use of that software unless I had some way to modify it and run my modification when I log on to the service. Not impossible, but not really any more practical than if I were to download the free server software, modify it to my needs, then install and run it on a server under my control.

    ReplyDelete
  6. What I was trying to describe was a kind of P2P cloud, distributed across donated machines - a bit like Fon, I suppose. *You'd* be running the software, not anyone else. The difference from running a server is that it wouldn't be tied down to one location, and it would have redundancy to cope with downtime.

    Any clearer? (Please bear in mind that I am making this up as I go along....)

    ReplyDelete
  7. Anonymous12:36 am

    Doesn't Amazon's EC2 service fit the bill here? (although I like your P2P idea as well)

    For those who aren't familiar with it, you basically get access to one or more virtual servers on a pay-as-you-go basis, so you can start up a bunch of servers, run what you like on them and pay only for the time you use.

    ReplyDelete
  8. Anonymous2:49 am

    Ahh, P2P! I think I'm following you now.

    I run a server at home, another one at work, and manage a rented server at a web-hosting service, all running various bits of software I wrote, so I guess I'm stuck in that paradigm and was having difficulty seeing beyond it.

    What you're suggesting could work. I'm imagining something like an application platform built on a cross between bittorrent and folding@home where program and data storage as well as program execution are distributed among the participating peers.

    ReplyDelete
  9. I think the problem with EC2 is that you're still dependent on a single supplier, which effectively has power of you. What I was trying to sketch was a way to remove such single choke points.

    ReplyDelete
  10. "a cross between bittorrent and folding@home where program and data storage as well as program execution are distributed among the participating peers."

    Yup, exactly. Now all we need to do is to find someone able to code it (the last time I programmed was in Fortran....)

    ReplyDelete
  11. Anonymous11:20 am

    > I think the problem with EC2 is that you're still dependent on a single supplier

    Maybe, up to a point, but bearing in mind that the apps running on EC2 are uploaded by you, there's no lock-in here - in theory you can always take your business to another supplier.

    Of course, your P2P alternative blows EC2 out of the water in principle. I would love to help out should anyone ever get this off the ground!

    ReplyDelete
  12. Well, as I've written, I'm the world's worst programmer....

    Any suggestions on how a project might get going around these ideas?

    ReplyDelete
  13. Anonymous5:46 pm

    Hmm, I wonder how much mileage there would be in simply running a BOINC server?

    I confess to not knowing much about BOINC, but surely it's not just useful for weighty scientific problems - could it not be used for a "compile my linux kernel" type project just as easily?

    I expect there are limitations to this approach, but it might be interesting to see just how far it would stretch.

    ReplyDelete
  14. You'd obviously need to add some superstructure, but it might be a useful starting point.

    ReplyDelete