Hacker News Reader

Comment by angryasian | Hacker News Reader

was auto grouping based on your friends information ever really that hard. You look at your friends info.. i see they have set location as "cityxyz" put them in list "cityxyz". These friends have employer "wxy" create list "wxy" and place people in this list. I mean with the amount of structured data that people provided on Facebook.. didn't seem like it was that complicated to me.

View on HN

what katango is undertaking is an extremely complex task. they're using machine learning and AI. what you are saying is just classifying people based on heuristics, not AI and this is not what katango is doing. Basically think of AI as putting a bunch of characteristics of each person in a 2 dimensional matrix and optimizing their characteristics based on AI algorithms.

I bet the reason why google bought them is to apply their work to google+ and to compete with Facebook, which is where FB is lacking.

well what you are describing is hardly what katango was doing in their product. You are probably describing their big vision, which all companies use to sell their product (Color described themselves as the all eyes watching tool from Batman movie, far from reality). They did the standard lists of , location, work , school etc. Nothing as abstract that requires the AI you are talking about.

How do you know how they came up with those lists by simply grouping them into location, work, etc?

Look at the lists closely. Not everyone who went to your school is included in those lists

What you are seeing is what they want to show you. The AI happening in the backend is really complex. But of course, you don't have to believe everything I say. You should check out the background of those 3 co-founders. They are all AI guys, and they are not likely to want to solve simple problems.

Reply on news.ycombinator.com