My Photo

Your email address:

Powered by FeedBlitz

April 2018

Sun Mon Tue Wed Thu Fri Sat
1 2 3 4 5 6 7
8 9 10 11 12 13 14
15 16 17 18 19 20 21
22 23 24 25 26 27 28
29 30          
Blog powered by Typepad

Become a Fan

« Responsible Innovation: Staying Engaged With the Privacy Community | Main | Stress Relaxing, Eight Sharks and Evading Surveillance Cameras in London »

June 01, 2006


Feed You can follow this conversation by subscribing to the comment feed for this post.

Bob Gourley


I continue to dig into issues like this, including reading as much research as I can coming out of some of the big schools, but I can't find anyone who puts this so well. So thanks. Please keep writing on these topics.

I have one somewhat minor comment, that applies to your concept of data finding data and relevance finding users. I've come to believe that one of the most important parts of any architecture that tries to do this is the human part. For at least our lifetime, the human brain is going to remain the greatest processor on earth and is very definitely going to be the best processor positioned to evaluate relevance in complex situations. Of course I'm just stating the obvious. But the not-so obvious thing that needs to be built into any architecture is feedback on relevance from the user. That puts the great brain processor in charge.

In your four use cases above, relevance feedback would be in the form of:
1) Happy sleeper likes hotel and tells front desk
2) Paying customers buy more books
3) Parents notify system of false alerts and also notify system of accurate data so the system becomes better trained.
4) Relevant discoveries are noted in ways that train the system to produce more. Irrelevant discoveries are ignored in ways that train the system to stop reporting them.

So I'm wondering, does the mantra now become:

"create systems where data finds the data, relevant information finds the user, and the user assesses relevance."


Jason Watkins

I've found your content over the years to be extremely interesting. I keep coming back to this one in particular.

I'm wondering how practical issues of software engineering are handled? Taking your first example, there might be a plethora of reasons that we'd want to reschedule the maid: for example if you'd just gotten room service in the last few minutes, or if the phone is active.

When data finds data, how can we avoid a combinatorial explosion of application code to handle all these discovery events?

The comments to this entry are closed.