Four quick notes on Havoc’s GUADEC keynote:
- I did the right thing by not going to GUADEC, but man, I wish I were at GUADEC.
- Havoc (and others saying the same thing) are completely right that we must move towards deep web integration. I have no strong opinion on whether online-desktop is the right technological approach for that, but we have to start somewhere. Moving the battle to a development platform based on an open, widely-implemented standard (HTML/JS/etc.) maximizes our strengths and minimizes the strengths of our competitors.
- Many web services are proprietary, but we can’t just twiddle our thumbs while the Apache/rails/django/etc. people work on this problem. Like free software in the 80s, we must realize that our options are (1) work with proprietary (web services|kernels) or (2) become completely irrelevant while we wait for free (web services|kernels) to be built. I know which one I’d rather do. As a bonus, I don’t believe we’ll have to wait for good free services nearly as long as we had to wait for a good free kernel.
- Hopefully I’ll get to do some interesting stuff with a Free Services Definition in the remaining weeks I have at Red Hat, now that GPL v3 work has mostly wrapped up. Am very excited about that.
9 thoughts on “four quick notes on Havoc’s keynote”
Sounds like you just predicted Alex’s keynote too :)
You should be here
[…] tree. Ria Das: Got new machine :) Eugene Teo: Singapore Linux Meetup Group – July Meetup Luis Villa: four quick notes on Havoc’s keynote Havoc Pennington: Keynote Reactions more Home » Fedora […]
I really don’t think that moving to an HTML-/JS-based development platform would do anything except slow the system down. You see how sluggish Firefox is? Now apply that to the whole platform. We’re better off making sure our C development tools are great, and adding functionality so that people can utilise web services, rather than converting the desktop to one.
I understand the comparison you’re making between web services and the proprietary kernels of yore. Unfortunately, there are complications that keep these from being completely analogous.
Terms of service for some of these web services are so strict, they’re completely inimical to freedom. Amazon’s has lots of rules about how you can download, display, and cache the content you get from them. I’m under the impression that others require that you not share your key — requiring every user to get their own, which may work for hackers but isn’t going to fly for most others. (Didn’t Deskbar do this for Google searches at some point? I thought it did but can’t find it now.)
On the other hand, it seems like things are definitely moving in the right direction. I recently read about one service that required you to give a valid URI as your key, and that URI was supposed to provide information about what the application was and did. So I think there may be ground to push for at least some minimum standard of freedom for web services that want to play with the free software world. Which is good, because I think that’s necessary for us — after all, every proprietary kernel of consequence let you run whatever you wanted on top of it, no matter what what the license.
[I admit I have not given this incredibly deep thought yet; this is mostly a gut feeling spurred primarily by watching our current clients becoming increasingly irrelevant to users. This irrelevance is bad in and of itself, but it is even worse once you consider that it weakens the open standards which protect our ability to participate in the broader data/software ecosystem, allowing them to be attacked by proprietary tools like Sharepoint and Exchange.]
People are actively choosing those TOSs, Brett, in vast numbers- in many cases faster than any other forms of software have been adopted before. Certainly faster than they’ve ever chosen any libre software before (with the potential exception of Firefox, which they use primarily to access… proprietary websites with evil TOSs.) So there is something there that we aren’t matching yet with libre software.
So, people are choosing non-free software (with as you point out even worse licenses than Windows) in vast numbers. We can sit here and twiddle our thumbs, and let them continue to use windows/apple to access those sites, or we can give them compelling reasons to use free clients, and then use the leverage of a free client to protect libre/open standards, to favor libre services when they are available, and otherwise Do Good.
Or to put it another way: what if those proprietary 80s operating systems *had* had incredibly restrictive TOSs? Would RMS have just magically pulled a kernel out of his ass? Would he have ceased distributing or working on the client tools? He might have put slightly higher priority on the kernel than he did, but he wouldn’t have stopped doing everything else; those other things were still necessary to form a complete ecosystem, and building them advanced freedom even if they weren’t complete solutions. Failing to build the other pieces while blocking on the existence of the kernel would have been stupid.
I’m not saying I like this; all things being equal I wish great Libre hosted web services were falling from the sky, and I’ll do my part to help that happen. But in the mean time, not having given it incredibly deep thought yet, pretending proprietary web services don’t exist seems like a recipe to quickly become even less relevant than we are now.
[…] I’m not entirely sure I like this progress but I have to agree with Havoc Pennington and Luis Villa: It is the future. Maybe not this implementation, but the idea… It will be very interesting […]
I was trying to draw a distinction between two kinds of TOSes in my comment, and in retrospect I don’t think I did a very good job of drawing the contrast. I was trying to draw a line between
a) web services with TOSes that have requirements which fundamentally interfere with the freeness of free software clients, and
b) web services with TOSes that don’t have those requirements.
I’ve only recently started thinking about how free software can , and/or ought to, interact with these services generally–basically when you mentioned it on a committee call a while back. Shortly after that, I received an e-mail from a developer who wanted to use Amazon’s web services to provide metadata for their music player application. I started reviewing their TOS, and it was not a pretty picture. It dictates very low-level behavior: you have to credit Amazon in very particular ways, when you download data from them you’re only allowed to cache it for a short length of time. Even if those sorts of things are the only problem, they threaten to undermine any free software client someone develops: your freedom to modify is significantly limited.
I still haven’t done enough thinking to have a reasoned opinion about how free software should interact with the second category of services. But I personally think we should take a stand against the first category. Yes, they’re popular. So are DVDs–but I don’t hear freedom fighters calling for their distributions to ship proprietary, DMCA-kosher DVD-playing software.
I understand your comparison to the proprietary kernels of the eighties, and it has merit, but I don’t think the situations are completely analogous. I’m more inclined to compare it to video cards. Today, we have complete freedom over our desktops; there are several in the FSF office with LinuxBIOS on the bottom and gNewSense above it. When we sacrifice that freedom to try to attract users to mostly-freedom, we risk losing it permanently. After all, despite years of effort and protest, neither ATI nor nVidia have released real free drivers for their hardware–and this even after ATI was acquired by AMD. It’s very hard to say what these sorts of compromises have gotten us.
And I personally feel a little optimistic when it comes to web services, because I think there’s a lot more potential to move people in a good direction here than there is with video cards or wireless drivers. It seems unlikely that patents or government regulations or other complications will muck up the decision-making process for the service providers: it’ll be purely about their self-interest. And who doesn’t like more users? Like I said, the trend seems to be positive; I’m under the impression that the TOSes for Web 2.0 services are generally saner and fall into the second category. There seems to be opportunities to pressure those in the first category without giving much up on our end.
[…] can never create the same level of pain that a proprietary OS can do.” Not true- explain why? TOS problem, and/or how we should use them Bailments FreeOpenServicesDefinition (last edited 2007-08-03 14:43:19 by […]
[…] The document has moved here. […]
Comments are closed.