Puffy and Chrome OS
About ten years ago, Tim O’Reilly made the claim that web apps circumvented copyleft licenses, since they were a distribution channel of sorts, but were not required to distribute the source of their applications. At this point, i had an iota of a notion.
1 Conversations about this phenomenon routinely involved the phrase “no evil” (which i find naive), and frequently devolved to the “convenience trumps everything” argument.
A few years later, Google released GMail (beta). I watched as dozens of my friends, who were diehard FLOSS folks, and who wouldn’t trust Microsoft, IBM, or Oracle with their desktops or servers, happily started trusting Google with their personal information and conversations1. At this point, i had a musing.
A couple years after that, i worked at a web application provider, where we handled and stored our end-users’ documents, and routinely had data loss problems. At this point, my musings started solidifying.
Then, last month, Opera announced Unite. And i realized i needed to get my idea out there. So i started talking to friends and colleagues, testing the waters. The name “Puffy, the personal cloud” first appeared.
Google is going to give you a free operating system designed to make it as easy as possible to part with your data.
Today, though, Google has Chrome OS in play. And that’s the big news, right? Bigger yet, in my opinion, is Google Apps leaving beta. Or at least bigger in conjunction. Why? Because Google is going to give you a free operating system designed to make it as easy as possible to part with your data. And apps which are out of beta can now be used by businesses. And businesses can be charged money. And people who use apps and an OS at work will be comfortable using them at home. And, eventually, they can be charged money, too.
I’m not going to get all conspiracy-theory here. Google Apps/OS isn’t really the problem, just the largest symptom. The problem is that we’re no longer in sole control of our data. That may be inevitable, but at least we should be able to trust the holders of our data. Any entity with a profit motive to abuse our trust should probably not be trusted. That includes Google, Flickr, twitter, and all the rest of the cloud apps.
2 Though i’m absolutely open to a better name. Something that expresses the decentralization would be nice.
Thus, Puffy2.
What’s the big idea?
Trust the people you actually trust. That’s the core. Chances are you know someone who can competently administer a server. And when i say “know”, i mean personally. A man or woman you are friends with, or a relative of, whom you trust to not open up your diary when you’re out of the room. That’s someone you actually trust, and who has incentive not to betray that trust.
What’s the big idea, technically?
In the opening stages, it’s pretty simple. We gather the FLOSS best-of-breed of the following web apps:
- Blogging
- Microblogging
- Mail client
- Photostreaming
- Wiki
- Bookmarking
- OpenID provider
And we package them up, pre-configured with a webserver and mailserver, along with some tools to perform simple user management across all the apps simultaneously. That’s Puffy, the personal cloud. You get your friend to install it on a server(or perhaps you’re that friend and you do it yourself). Then you get an account on that Puffy. So do a couple dozen of your friends. With a low user base, the scalability of Google is not necessary, yet the convenience of an always-on, always-available app suite is still present.
That seems simplistic
It’s just a start. From there we need to ensure that Puffy apps are OpenID interoperable (so i can comment on my friend’s blog and photos even though she’s on a different Puffy). We need to truly unify the user management, including permission systems. We need to offer data imports from existing popular web apps. We need to bring in new types of applications. We need to offer options for the customizers out there.
And, perhaps most importantly, we need to ensure that these apps are well-integrated with mobile devices and netbooks, as those are the places they’ll most be needed.
We can do it, though. Chrome OS won’t arrive till second half of 2010. That gives us a year. Who wants to help? Even a suggestion of a best-of-breed app would be a step in the right direction.
By john on July 8, 2009
Comments
By ozten @ 4:11 p.m.
ozten: You’re right, there are definitely problems (or even tradeoffs) involved. Specifically, there’s the aggregation problem you mention, but there’s also the privileged information problem (e.g. LiveJournal doesn’t just aggregate your friends’ posts, it also allows you to read locked posts if you’re authorized).
The aggregation problem can be mediated by integrating federated search. There’s 6earch, for example. Labelling other Puffys as search peers would allow your ‘cat’ search to cascade out in several directions. This isn’t a solution, exactly, but i’m not sure one exists—ultimately, in a decentralized world, the notion of a centralized list becomes less feasible.
The privilege problem is more difficult when dealing with the legacy cloud apps, though there are some which allow for OpenID authentication. I’m not sure what the solution is, there.
When it comes down to it, though, data is the currency of the next decade. There is great incentive to take it from you, and the people who have the technical knowhow to effectively keep their own data are in the minority. Puffy lets those people serve their friends, and keeps any single entity from having a strong data monopoly.
By john @ 4:44 p.m.
Comments currently disabled due to impressive comment-spam efforts.
Interesting post!
Is recreating and hosting social networking sites the right strategy?
The utility of many of these apps comes from your social graph. If you have many puffies, each with 10 users, then you lose this ability.
The way I use delicious is to harvest new links on a topic. This wouldn’t work if I used my open hosted bookmark service with 10 users.
Another strategy is for Puffy to be a import/export/sync tool for the best of breed apps. This gives users control over their information in some powerful ways, which they don’t have today. It doesn’t do anything for privacy concerns however.
F/OSS software protected word processing programs software by creating a F/OSS word processor.
It’s possible that the analogy carries to “cloud” apps, but I’m thinking it’s a different beast. Say we clone Flickr and you have 500 instances running with 100 users each… how do you exchange data between instances so that I can look at all photos tagged with ‘cat’?