In his EconTalk podcast, Hal Varian, Chief Economist with Google speaks about Search amongst other things.
..search is about two things. Precision and Recall. He then defines precision as “off the number of documents retrieved by a search result, how many are relevant” and recall as “off all the relevant results, how many were retrieved”.
This in my opinion is very important aspect. Before we take on a problem to solve, we need to have the exact definitions in place.
He also spoke about co-evolution of technology and users. He said that any technology and its adopters evolve rapidly with time. He took examples of cars and extended it to search engines.
..users are getting more skilled at entering search queries. This helps both user and the search engine in delivering results that are high on precision and recall. And thus making the experience better for the user.
This probably explains why most of the user-dependent (user generated, network effect etc) businesses remain in beta stages for long times (gmail is still in beta). Till the time both (business and users) evolve and are mature enough to help each other, system has to remain in beta.
Lessons? Help your users evolve. Give them tools of the trade. They would need some hand-holding to learn initially. And then they would need the freedom and independence to play around with the system before they start making effective use of the system. And same goes for the system as well.
Please note that the definitions might not be exact (since I took notes while listening to the podcast), but the essence has been preserved.
- Link to podcast.
- Image via Wikipedia
- Originally posted on Saurabh Garg on New New Thing
Thanks for voice to txt. Liked it.
co-evolution(of users and system) is not the best strategy. We might get struck in a local maximum. We’ll be tuned to think in keywords instead of thinking in our natural language. Communicating in keywords is a lossy compression of intent!!
But yes, keywords do save a bit on typing and remove redundancy. I’m not sure if thats a good trade-off.
-Mahesh
Celebrating Life…
Hello everyone
I think there are enough entrepreneurs who frequent this forum so I would need your help in resolving our problem . we r 3 girls Tania, Nherlyn and Nancy who maintain this hugely popular blog. we get around 10000 hits per post and we think we can slightly monetize the traffic. but we have no clue how to do it. can any of you give us some tips or ideas..
@Mahesh T .. Why do you think co-evolution is not desirable? I think its the best insight that someone could have given..
Can you please elaborate on local maximum .. ?
@Tania .. spam .. ?
Regards,
SG
@saurabh Yes, co-evolution of users and search engines is desirable but it may not lead to THE best outcome. Will we start to think in the constraints imposed by keywords instead of rich natural language queries?
Evolution with one variable fixed brings out the best(global maxima); We can imagine what the best looks like. But, co-evolution might lead to a local maxima(instead of the global maxima) because, at each step we are trying to solve the immediate problem ahead.
If Tigers and Humans co-evolved; we might have become better runners. We wouldn’t have built a spear to protect ourselves.
-Mahesh
Celebrating Life…
@saurabh I have been thinking about co-evolution of technology and users. You were right and I realized this over time, its actually great insight. Its the right thing todo in the real-world. This(keyword search) is our best tool given where we stand(NLP technologies). It doesn’t matter if co-evolution was THE best strategy looking backward. Time is irreversible and we make the best out of what we have
-Mahesh
Celebrating Life…
Hi Mahesh,
Glad I could be of help.
Regards,
Sg