Welcome to WebmasterWorld Guest from 18.104.22.168
Forum Moderators: martinibuster
I've no doubt that MyGoogle will follow soon and MSN are already personlising via messengerclient/webmessenger/hotmail homepages.
I'm just pleased I am no longer having to try and provide an optimisation service based on what's happening in the marketplace.
gmail interface messages, they only take up a line, until you click them and they instantly expand.
Now think of SERPs. Each site gets a spam probability rating according to some sort of spam algo. If the rating is above a certain point, it doesn't get removed, it gets colalpsed to single thin line with the title and the spam rating displayed. For every 2 or 3 collapsed SERPs, an extra item is shown on the page, to keep it approximately the same pysical length. Now people can judge by titles if teh spam is worth a look. Algo can be tweaked to false positive markers, i.e. people visiting sites with high spam rankings.
PM me for royalty payments ;)
What we need to be doing instead is to visualize the most desirable situation and work on progressing towards this vision, considering and accepting today’s solutions as logical trends and a progression!
Even if this new solution can't address that future vision as an end result, we need to ask ourselves: is it a stepping-stone and will it help to reach the next level(s).
skibum said: "Presumably, most of the time you go to a search engine to find something new, not to find your favorites?..."
Others claim they would not jump back and forth to set filters, bookmarks, blocks and the likes.
Assuming this is a step into a future direction, combine it with speech recognition, profiled and dynamically recognized preferences and you made a quantum leap!
Semel for president!
Y! could NEVER control the abuse at that level and they do know that, I am sure.
or one step further --- if the search engine could effectively PREDICT that I don't like these types of sites and rank them lower
In my mind, the above would be a part of search actually becoming personalized. However, I'd say it's further than a step away. It's one thing to let Joe Blow tell you he doesn't like Site A, but it's an entirely different thing to know he wouldn't like Site B. There are just too many variables in place. What exactly did the user not like about Site A?
# high prices
# garish colors
# high shipping costs
# affiliate content
The list is endless.
Thanks to the domain name resolving question, and the Google API, I have actually set up a site to do something along these lines. You can see the results in my profile (I hope this is ok - completely non-commercial site, proof of concept).
The idea is simple - given a search result set, select the results that aren't what you are looking for and the search engine refines the search based on the sites you tell it are not valid.
Ok, so I don't have this set up as a big huge user database that stores permanent information, but thats more of a resources problem that software or technology. This was actually done last evening - starting from scratch. I don't think what yahoo is doing is that terribly impressive as it stands. Things like this concept in my profile - where the search "data mines" my preferences to retrieve new results, are a bit closer to a search engine that knows exactly what answer you want.
By the way, if you do check out the site in my profile - I suggest glancing over the usage instructions, specifically the example of a search for "Jaguar", an ambiguous term.
The future looks bright ... where are my shades?
BTW, their "launch" radio station works along similar lines, and that works like a charm. Each time you hear a song, you can rate it with 1-4 stars or block it from ever being played again. You can also rate artists, albums, or genres. After some time, it knows your taste, and you will never again have to listen to Billy Ray Cyrus :-)
Your preferences can be shared with your friends. New songs are played according to a "you liked that song, then you probably like this song too" or "people who have rated Radiohead with 4 stars like this song" etc. I'm sure they are trying to reproduce this system to some extent with tis mysearch thing...