One size fits all, badly
Specialised search solutions, means more feature rich search solutions. This has been true all the time, but connecting to the data (crawling) has been a lot of work historically. With better APIs and more services actually having an API, this is changing. For my pet project: life-index, I’m using IFTTT to connect to my data sources. To gather the data, not a single line of code is used. Google was great when it came, but one answer to every (search) question in the world?
Crap to content ratio
A really tiny fraction of what Google indexes, is of any interest to you. No matter how good Google’s relevance model is, there will always be a lot of uninteresting stuff served by Google Search. More specialised search solutions won’t have this problem. I.e. a recipe search.
Natural scalability and advertisement income to hardware cost ratio
With smaller indexes and a search engine running in the browser, you can move over to users hardware. This has been true for quite a while, and now also for search engines with search-index. This makes the advertisement income to hardware costs ratio higher. Google, with it’s large search index, can’t do this.
Mats Olsen explains search-index and natural scalability.
The logged in web
A lot of services that demands login to post and read, is naturally not indexed by Google.
… and why this isn’t dooms day for Google
Google will adapt. And when they don’t do it quickly enough, they buy companies that do. This is not about Google’s downfall, but how what they do now, may not be the right thing in the near future. And of course, Google is fighting a lot of other battles, and doing quite well.
So, is Facebook search, Twitter search and Gmail search the solution?
Sometimes, yes, and often no. Sometimes, they are better than Google within their own domain. But I’m betting that no is the answer with my personal search prototype: life-index. I was using 60+ online services last time I counted. Services where I have to log in. Maybe 1/3 or half of them have content that could be interesting for me to search in. But figuring out where to search is an analog search job in it self. That’s why I want all “my” content in one search solution.
So, what do you think? Am I wrong, and if so, why?