November 4, 2005
@ 06:09 PM

I saw someone reference the Dave Luebbert's reasons to clone Google's API and wonder what my opinion was in response. In my post from yesterday entitled Clone the Google APIs: Kill That Noise, I gave some technical reasons why we wouldn't want to clone the Google APIs for Windows Live Search.

However, there is probably a clarification that I should have made. In certain cases, there is one thing that trumps all technical arguments against cloning an API. That is when the API has significant market share amongst developers. This is one of the reasons why even though I thought that the MetaWeblog API is a disaster, we made the call that MSN Spaces will support the MetaWeblog API. Since the MetaWeblog API is a derivative of the Blogger API, you could argue that in this case we are cloning a Google API.

To me, the difference here is the case of mindshare. The Blogger & MetaWeblog APIs are widely supported across the weblogging industry and have become de facto industry standards. I don't believe the same can be said for the Google's search API. If anything, I'd say the OpenSearch is the closest to a de facto industry standard for search APIs although [for now] it has been ignored by the big three major search players.

On a similar note, I'd probably agree that the Google Maps API is probably on its way to reaching de facto standard and Yahoo! & Microsoft should just go ahead and clone it. If I worked on the mapping API for either company, I'd probably give it six months and if adoption hadn't increased significantly would consider cloning their API.


 

Friday, 04 November 2005 20:56:16 (GMT Standard Time, UTC+00:00)
Dare -- very valid points. And I'm glad you brought up OpenSearch. I just posted my thoughts on open search APIs. Here's the link (please feel free to reformat if necessary):

http://www.unto.net/unto/work/on-open-search-apis/

Specifcally, I address why I didn't clone someone else's search API when creating OpenSearch at A9.

Cheers,

-DeWitt
Friday, 04 November 2005 21:06:48 (GMT Standard Time, UTC+00:00)
While the big 3 haven't yet adopted OpenSearch (although all three have RSS or Atom search results for at least something), I would almost go as far as to say it doesn't matter - the reason to have tons of search engines is because all the niche ones can do better than the general ones, *because* they are specialized.

That being said... Google, Yahoo, MSN... come join the OpenSearch party :-)
Monday, 07 November 2005 02:44:17 (GMT Standard Time, UTC+00:00)
As someone who has written a library to abstract out the differences between search engine APIs (see http://argos.dev.java.net/) I can say with a little authority that just cloning the Google API isn't the way to go. As you said, using SOAP is the biggest problem (although in Google's defence there I think there are some historical reasons why using it made sense at the time - it is a lot older than the other APIs).

REST+RSS/Atom makes the most sense to me. OpenSearch is a great way of standardising that.
Comments are closed.