Google Base basics
Alright. I've had enough questions related to Google Base lately that I wanted to add my two cents beyond the post below. The 'secret sauce' in Google Base is tied very closely to the folksonomy they enable to their listings. In my humble opinion, Froogle failed because it was a centralized attempt to organize a massive amount of information into a product search experience. Froogle managers essentially have to define for each product out of millions (or worse yet for each keyword out of billions) what attributes are usually used by buyers for narrowing their product search. So for an Ipod that might mean 1) color 2) memory 3) type (mini, regular, nano) but for a Kodak digital camera those attributes would be completely different. The easy universal attributes Froogle has applied currently are things like 'price' and maybe 'brand', but ultimately something that basic is set up to fail. Buyers need more help than that...
Google Base takes a completely different take on this. Users set their own taxonomy of attributes and categories. As more users add content to Google Base, the overall number of attributes users select for any one keyword or product increases as well. All Google Base needs to do then is apply a bell curve to measure the most popular attributes or categories selected for said keyword search term and tada! you have a universal attribute structure which constantly adjusts and improves based on what the community of users generate. At first this process is a complete mess, and is hugely open to abuse by spammers, since the 'folksonomy' doesn't have sufficient number of attributes inputted (hence my search result for 'recipes' turning up jobs and job type as attributes below - this is already gone btw replaced by the more appropriate cuisine and ingredients). But as 1 million listings turns into 100 million (so long as its not all spam) the system essentially improves itself, and the most popular attributes become the most appropriate for display to buyers. Google rids itself of the work required to select attributes product by product since the user community does the work for them... Go spam go!
Update: Here's a story about the lagging Froogle. Just in case you didn't believe me ;)
Google Base takes a completely different take on this. Users set their own taxonomy of attributes and categories. As more users add content to Google Base, the overall number of attributes users select for any one keyword or product increases as well. All Google Base needs to do then is apply a bell curve to measure the most popular attributes or categories selected for said keyword search term and tada! you have a universal attribute structure which constantly adjusts and improves based on what the community of users generate. At first this process is a complete mess, and is hugely open to abuse by spammers, since the 'folksonomy' doesn't have sufficient number of attributes inputted (hence my search result for 'recipes' turning up jobs and job type as attributes below - this is already gone btw replaced by the more appropriate cuisine and ingredients). But as 1 million listings turns into 100 million (so long as its not all spam) the system essentially improves itself, and the most popular attributes become the most appropriate for display to buyers. Google rids itself of the work required to select attributes product by product since the user community does the work for them... Go spam go!
Update: Here's a story about the lagging Froogle. Just in case you didn't believe me ;)
posted by Ro at 11/17/2005
3 Comments:
Then, once the database becomes more structured with appropriate attributes for each of the categories and items, Google will slap on a really nice search, browse, and transaction interface. Then, they'll roll out the Google payment system, which, if used by buyers/sellers of Google Base items will offer fraud protection and a reputation system.
All this without the pesky listing and FVF fees on eBay, for low, low price of just a 1-2% transaction fee.
At which point eBay will recognize the competitive threat, and regret the fact that they didn't start innovating a lot sooner.
Well... free sites, like free email, is a ripe area for spam abuse. Won't help to have 1 billion spam listings vs 1 million real listings and try defining attributes that way.
Don't forget...people are basically good... :)
Post a Comment
<< Home