Hacker News Reader

A thought: one way to do this might be having a service diffing the HTML of whatever page guide X is for. For things like Facebook your diff would have to be smart enough to differentiate between 'content' and 'structure'. You could then flag guides based on that. But I imagine most guides are written/maintained in-house so this isn't a problem at this point.

Where do you guys work from?

View on HN

We collect lots of information about the element that we need to highlight (text, depth, siblings etc). When we fail to identify element accurately, we try to find best one with highest score and highlight. We mark flow as a failure and exit, when we hit score below minimum. This algorithm is most crucial one & i guess it will be evolving ever. By this mechanism, we are trying to overcome page changes to good extent.

Very few guides are in house and mostly are contributed by early adopters. we are expecting users to contribute.

We are based out of bangalore, india.

Reply on news.ycombinator.com