On 27/11/2007, Brian Butterworth <[EMAIL PROTECTED]> wrote: > > > > Sure, and I'm suggesting that a common API will be a base that each > > gatekeeper will add bespoke features too. I'll be surprised if similar > > services offered with a "common open API" from Google and Yahoo and > > Microsoft do not have any specialist features to differentiate them. > > Does the track record of Microsoft not show that they created aparent APIs > and then did not use them themselves?
Right - this is the problem with saying "Oh, I _could_ download the source code that runs this API interface, but actually I'm not going to bother and just use the service provided." - there is no guarantee that what you can download is what is running on their server. I'm lumping Google, Yahoo and Microsoft together as all equivalently bad - along with all smaller "Web 2.0" API providers who provide computation services through web APIs, making a subtle distinction between those servces and those that are data request protocols more sophsticated than HTTP. "Use our stuff to make your stuff," to me, refers to providing BBC data for the British public to do computation with, and where Backstage publishes APIs that take input data, transform that data, and return the result, the source code for the transformations ought to be published as free software - preferably under the GNU Affero GPLv3 or some other network-aware copyleft license. > > > > Freedom means more than a choice of lords. > > > > > > You can happily run your own things and then be your own lord, > > > > ...but not if the gatekeepers continue to offer software to the public > > without making the source code to that software public. > > To be fair, I think that there is an important point missing here. > > A closed API is fine if the service is best run somewhere remote. Well, if Google is searching _their_ data - their copy of the web, say - and returning the results to you, that's good. Its their copy, afterall. But if you upload your data - say, your spreadsheet numbers and equations - and get them to do your computation for you, instead of using Gnumeric or OpenOffice, that's not good. > For > example, if I need to use a database engine, I'm really bothered about how > well the API supports my SQL statements, not how they are executed. Right - because SQL is a "query language" for requesting data to be sent to you, for you to do your own computation on. > On the other hand, software that is transferred onto my own machine, I care > more about. I seem to have a concept of personal rights that extends to my > computer's CPU. If the code is in my machine, I should be able to know how > it works, if I so choose. Right > It's a strange concept though. The logic of the argument is that I could, > if I had the time, work out how it all works from the assembler code. http://en.wikipedia.org/wiki/Reductio_ad_absurdum > If it > can be run on my computer, then it has to be in a published format. CPUs > can't be "closed" because if the manufacturer refused to let you know the > instruction set, they wouldn't sell that many. > > But is that an API? Where is the boundary between CPU and APIs? Source code. > At the other level (and bringing it back to backstage) what about the well > known API of RSS? Should I care how the RSS feed is created? Does the RSS feed contain the BBCs news? Or does it contain your data that has been transformed in some way? -- Regards, Dave - Sent via the backstage.bbc.co.uk discussion group. To unsubscribe, please visit http://backstage.bbc.co.uk/archives/2005/01/mailing_list.html. Unofficial list archive: http://www.mail-archive.com/[email protected]/
