mca blog

the recent kerfuffle over the possibility that delicious.com might close should be seen as a clear reminder of the harmful detour the Web has taken over the last several years. there is noticeable sturn und drang over the matter, but - i suspect - for all the wrong reasons.

no doubt, it is quite troublesome for users to learn a service they have grown to love, possibly depend upon, is going away. however, the problem(as i see it) is not that someone has decided to close the service that is at the heart of the trouble. instead, the real problem is that the data user place in the system is either lost or difficult to recover. and even in cases where the data is recoverable, it is usually in a form that is less "use-able" than when it was hosted by the service and often in a form that makes importing that data into a similar service difficult or impossible.

it's the data, stupid

the problem is not centered on the service, but on the data. most of us have been seduced into storing our data in locations that we, ourselves, do not control. along the way, we are enriching others in the process (the service providers) and, when the well runs dry, usually getting little to show for our generousity.

even worse, we suffer the possibliity that we will loose the very data we we storing and have little recourse.

OPD - Other People's Data

of course, this is not a new issue at all. the most recent round of "Web 2.0" ventures have relied on the idea of using other people's data in order to gain a profit. in return, users are granted some "utility" that makes it worth while to deposit their data with that service. this is a modern variation of "OPM" (Other People's Money); the notion of using the resources of others as a way to make your own riches. in 1914, it was all about the bankers. recent events in the financial sector lead me to believe that 100 year-old lesson still hasn't taken root. i suspect the same might be true for the information sector.

"i told ya, it's MY data!"

there are ways to provide the same "utility" to your information w/o requiring you to store it somewhere else. for example, use a URL-shortening service (Adjix) that, as a feature, writes all my data to a storage provider of my choosing (i use Amazon S3 for this right now). if this fine service ever shuts down, i won't need to "export" my data - i have always had it; readily available. it's only the "utility" of generating short links that i will loose if the site shuts down.

see, this company provides me a service, but they don't keep my data. and almost all "Web 2.0" sites can do the same. if they are designed properly.

the Web of Services?

unfortunately, the current crop of web sites (web services, web applications, etc.) are not designed properly. they assume that all their data is local and that they have direct access to that data at all times. this works for implementations where the data is the property of those who operate the site, but it's a flawed model for many sites that operate on other people's data (OPD).

mostly, this shortcoming occurs because those implementing these sites are using data (and networking) models that make sense for local networks, but are counter-productive for distributed networks. we've all seen this before. complicated access controls to support multiple users in the same data system where the data must be kept separated. mind-boggling transaction management details in an attempt to support data integrity when multiple users could be writing multiple records from multiple locations for multiple partitioned data segments, etc. spaghetti-like UI coding to support unqiue user customizations of display all from the same code base. the list goes on.

the result is that we've designed a web of service applications instead of a web of data. we've implemented hundreds of data islands w/ only a tiny bit of interaction and co-operation. and it's sure to get worse before it gets better.

the solution is simple (and radical)

if we want to reduce the nubmers of data islands, if we want to limit disruption when sites close down, if we want to improve the possibility of co-operation among sites on the Web, we need to do a very simple thing. we need to take back our data.

and the easiest way to do that is only use services that let us decide where the data is stored.

that means applications on the Web need to be designed and implemented differently. they need to focus on offering utility and on doing that without holding our data hostage. they need to be real Web sites.

while it's a simple idea, it is also rather radical. it would force most of the current crop of "2.0" sites to fundamentally change not only their data model and their application model, but also (most likely) their business model. it would force these sites to ask for out permission to use the data. it would force them to focus on what value they provide to users, not to advertisers. it would put users in charge of their data in a fundamental way.

don't get caught "Gawker-ing"

even though this is may seem to be a radical idea, there is clear evidence that the primary notion behind it is taking hold in one area in particular; identity. the recent mess w/ Gawker.com sites remind us that centralizing identity (in this case using Facebook, Twitter, etc. to authenticate to a site) protects us from situations where the data we provide to these Web sites is mis-handled in some way.

it's worth noting that, while keeping the identity information out of the target web site's storage is a plus, placing that same identify information in the hands of some other third party (e.g. Facebook, Google, Yahoo, MSN, etc.) is much akin to re-arranging the deck chairs on the Titanic. there are efforts underway to make identity truely distributed (for example WebID). time will tell how that works out.

So...

so what's the point of this ramble? it's just this:

as long as we continue to build and support web sites that hold our data, we'll be ready victims of data abuse|loss.

just as the common advice today is for new start-ups to leaverage external services for user identity, my advice is that these same sites should also be designed to support external service for data storage. IOW, be prepared for users to want you to store their data in user-designated locations. i know it's taken years for third-party authentication to gain momentum, it will proly take years for third-party storage to do the same. but, in the long run this is the proper model for the Web.

i, myself, continue to look for sites that allow me to controls my own data storage and select them whenever possible. as more of us do the same, it's likely more sites that support this feature will emerge. and, eventually (hopefully) we'll reduce the distruption caused when sites get shuttered.

who knows, maybe the idea of "storage-abstraction" is the starting point for "Web 3.0"