Search has become synonymous with Google and now Google has decided to share its search algorithm updates on a monthly basis. However this new spirit of openness isn't real - Google still depends on security by obscurity to stop people from manipulating its results.
The people responsible for Google Search are very keen for us to know what they are doing.
A video posted earlier this year that we commented on in How does Google search?, tells how there are around 500 changes to the search algorithm every year - that's around two per day - and discusses the process of making these changes testing them to ensures that they are perceived as improvements:
In November a post on Inside Search, the official Google Search blog, listed 10 algorithm changes and as that was well received it's becoming a regular feature with monthly lists of changes.
Top result selection code rewrite: Avoids “host crowding”, show too many results from a single site. Code is easier to understand, simpler to maintain and more flexible for future extensions.
More comprehensive indexing: Makes more long-tail documents available in Google's index, so they are more likely to rank for relevant queries.
More autocomplete predictions: Makes the prediction algorithm a little more flexible for certain queries.
Original content: New signals added to make better predictions about which of two similar web pages is the original one.
Live results for Major League Soccer and the Canadian Football League: Displays the latest scores & schedules with quick access to game recaps and box scores.
If you are interested in the Evolution of Google Search this video posted last month shows some of the notable milestones and introduces more members of the Search team.
Google has an interesting problem to solve. If it opens up on its search algorithm then people will find ways to manipulate it. However if it says nothing then users might lose confidence in it because it is opaque and not trustworthy. Saying just enough to indicate that you are ahead of the game but not enough to enable others to win the game is a difficult problem. This is security by obscurity and we all know that this is never a good idea. Google really should put some of its research efforts into finding ways of basing its search on methods that can be made public without too much danger of being manipulated.
Given the controversy surrounding chatbots and the Turing Test, it would seem unwise for neural networks to challenge the same problem. As you might guess, they have and the result is the predictable [ ... ]