The Burgeoning Openly Owned Web

So here’s an idea that Dave and I have formed over a late night of brain-storming - it’s a work in progress and items still to be fleshed out - read and tell us what u think!

The Open Web Manifesto (Draft)

• We believe that we should be able to CRUD our own data on our own terms
• We believe that our data should be owned and accessed directly through open standards
• Proprietary web services take our rights to our own data away
• Portability is our right

The disintermediation of personal data
Our personal data is increasingly entrusted to third parties, for instance:

• Your Photos reside in a proprietary File system (Flickr, Picasa etc.)
• Your Email resides in a proprietary File system (Gmail, Yahoo etc.)

This is a bad idea, because you have no control over your own data. Your data is stored in proprietary storage, and is hidden be hind proprietary web applications. Some of your data is accessible over the published API’s, but you have only limited functionality, and you do not have any guarantees over the long term life of your data.

Taking Back Control
We think that people are going to start wanting more and more control over their data. Services like Amazon S3 and Dreamhost provide inexpensive, enterprise class data storage. If we can access many (or one!) data stores in a useful way, there is no reason why we should use other proprietary web ‘land grabbers’. You can easily move all your data to a secure, agnostic online drive with multiple backups over multiple vendors if you so desire. Or, you could keep a backup of your online data at home on a USB drive. It’s YOUR choice.

Costs

• The bad news is that you’re going to have to pay to host your data. But from there, it’s all free (if you want it to be).
• The good news is that utility storage services (e.g. S3) charges a lot less for data storage than Flickr! And by choosing your storage provider you can store files up to what ever size you want!

Accessing your data

• We think that you should decide how your data is accessed, and who gets to access it.
• We want to develop open source services and tools to assist you in accessing your data. Web services are analogous to software, in that they should be free and open source. Although there is a cost associated with, say, indexing and then making your data available on a web service, we believe that these costs are very small and can be absorbed through donations etc.
• Creating an open market with low barriers to entry for, say, photo sharing apps (running on our aforementioned free web service) will drive down costs and promote better products.

Service need to be built

• A service which indexes you public and private data, and provides a ‘private search engine’
• A service which can display media (photos, video, music, documents) in a cool way
• A service that will sync your media to other services (flickrsync, picasasync)
• A service that provides a single interface for multiple front and back ends
• A service that manages your metadata (tags, EXIF etc)

What already exists
Tools to upload and download to utility storage services (e.g. S3)

Use Cases

• Dave lost all his old emails when he moved from hotmail to Gmail. If he had used S3 for his email storage, he could have backed them up, or just switched frint-ends.
• Andy uses flickr to store his photos. Most of his friends use Picasa. Andy can’t participate in their ’social web’. Andy should have used S3 to store his data, so he could then sync it using 3rd party tools with any online photo provider.

Draft
So I’ve been using EC2 for the last few days and while it’s a super idea in principle it’s pricing model is just a little unfair. Amazon say that per instance hour you will be charged $0.10 - great! But wait… even when your virtual instance is running at 98, 99% idle you are still being charged A quick calculation and using EC2, for just computing power, is going to cost $67.20 per instance…. for potentially doing nothing!
Yes I know an idle machine is taking resources, but surely the guys can come up with more innovative things? Dynamic and predictive computing resource allocation anyone? Seems others are beginning to wake from dreams too

So google look to be bringing such applications to the masses, Scoble has more details and O’Reilly too. Yea there have been notions of this in the more tech oriented press… Firefox 3 to add support, Dojo already implementing it in their toolkit and this web site claiming that they have it already in their web app. I’ve a feeling, that with this plugin, google will reach out to the masses and with their want on standardization this whole venture is very worthy. All we want now is for them to go GWT style shouldn’t be long Gears is open source and licensed under the New BSD license. and release it as open source so all can benefit from this goodness. They guys that implement this gotta get talking ASAP to the people responsible for GWT - seems like they already have and there’s code! There’s already a posting up on the Gears group. O’Reilly picks up that there is no real mention of how synchronisation will be performed and looking at the supplied code there isn’t many hint there either. By the way, my reader of choice, google reader works quite well with gears!
Fabric Live 23 Death In Vegas from the album “Fabric Live 23 Death In Vegas” by Death In Vegas

So I’ve decided to go with EC2 - it’s the future IMHO… utility computing and all that jazz. More compelling are Amazon’s revised costing structures.
How to go about using it? Sign up! If you are out of luck and can’t get an account rightscale can help you out there by acting as a proxy to amazon for you - bloody brill service. Excellent demo of automatic horizontal scaling by rightscale - I’d have questions on how automatic this is and if it can do adaptive and predictive scaling.
The documentation at the dev site is great so no need to repeat.
There’s plenty of us out there that wanna use the olde bauld ubuntu - there’s an image of it as a shared amazon public image - just one caveat - there’s no local loopback adapter configured so all you need to do is follow the instructions here. If that doesn’t suit then get familiar with debootstrap and read through here - it’s not alot! Some more resources describing the creation of custom AMIs can be found here and here
With this image up and instantiated, for my needs (at least right now), all I have to do is to install Tomcat (good resource here) and re-jigger the config file to listen on port 80 (yea yea I know - I’ll be proxying to it, but later).

Just some random stuff:

• Reading this for a nice and quick overview of REST
• Now here’s a TOC that I’ve rarely (never) seen!
• Type faces are as always important
• Power consumption is always a increasing issue - slipping from data centres to the home
• Idea: Wiki’s for contract negotiation - legal, non-functional or functional interfaces
• From CRN: “The concept of an activity is the idea that wherever I am, I can spawn an activity, create a team and people can ontribute work into that activity regardless of the tool I’m in or they’re in. It gives us a way to transcend the different stove-piped tools,” Bisconti said.
• Strategy is differentiated from tactical or immediate actions as being the magnitude of time frame and scope
• HOWTO: Adjust a sturmey archer 5 speed internal hub <- TODO
• So according to a myers briggs test I’m executive

What would be good to do with Soapod is to combine
the notion of a message router and application server - then u can do great stuff like URI based service lookup - essentially the node routes traffic a la IP but deals with the application request should the service exist on the node.

But what about the case of many services of the same type existing? In this case the node reporting the many services returns the list to the client which chooses wither manually or using a recommender based on the original client’s local knowledge/preferences. This could be negated by the selection of appropriate URI naming schemes if tagged based invocation requests did not suffice.
Perhaps a good vehicle to explore this idea would be Mule - an OSS ESB. What would be required is a dynamic registry of registered services (did somebody say OSGi? :-D) and their corresponding URIs. This OSGi service registry would ideal back end on to a folksonomic DHT service registry - a service already implemented in Soapod.

Expand: Enterprise mashups - resource oriented computing - services are queryable by RSS
e.g. a registry service:

http://host/registryService?listServices

returns a list of services available in the repository in RSS format

http://host/registryService?listServices&Tag=Photo

returns a list of photo services available in the repository in RSS format

Tokyo Black Star - Blade Dancer (Dixon Edit) from the album “Fabric 25 Carl Craig” by Carl Craig