Wow, this article demonstrates sophistry: "big sounding" but ridiculous ideas in regards to the scalability challenges involved in search (has the author ever heard of latency?), no mention whatsoever of the science-intensive nature of search (page rank is a beatifull algorithm that captures the imagination, but that's like narrowing down a commercial RDBMS to a B-Tree: there is far more behind the scenes that requires hundreds if not thousands of domain experts to research, implement and deploy).
This also missed the key thing Google has done: they're improved upon Overture's "pay for performance" model by incorporating click through rates into rankings which (when coupled to a scalable architecture that Overture didn't have until take over by Yahoo and Project Panama) turned them from "just another search engine" (I still remember our neighbours, both of them engineering managers in enterprise software companies reacting with utmost disgust to Google's IPO price: "they're only a search engine!") to the first company to make search hugely profitable to advertisers.
The "next Google" will take yet another commodity field that's ignored (has no clear winner at this time, not even seen as particularly technologically challenging: remember, inverted indices have existed long before Google) and make it a vastly profitable niche, while at the mean time "sucking up" Silicon Valley talent as to create a "moat" around their business. The talent could then yield secondary moats, much like Google's expertise at scaling out on commodity hardware enabled them to create the scalability moat at a time when all other players ran on mainframe hardware (with the notable exception of Inktomi, which started on clusters of bottom-to-mid Sun SPARC hardware, the kind that developers had on their desktops rather than enterprise and mainframe class gear -- quite close to commodity hardware), afraid to tackle the tough distributed systems problems.
I figured that by the title, which is why I checked the comments first. 'The google killer no one talks about' - my first thought - if no one talks about it it's likely not good and if no one is talking about it no one will hear of it and it'll 'kill' nothing.
Thank you gokhan, you likely saved me 10 minutes of a boring hyperbolic read. +1
Many years ago I thought gigablast was going to be the next thing to get close to the big three but now it's faded away. It really was impressive years ago, all coded by one guy and ran on like 5 machines.
Now it's turned to greenwashing which is strange to see.
This also missed the key thing Google has done: they're improved upon Overture's "pay for performance" model by incorporating click through rates into rankings which (when coupled to a scalable architecture that Overture didn't have until take over by Yahoo and Project Panama) turned them from "just another search engine" (I still remember our neighbours, both of them engineering managers in enterprise software companies reacting with utmost disgust to Google's IPO price: "they're only a search engine!") to the first company to make search hugely profitable to advertisers.
The "next Google" will take yet another commodity field that's ignored (has no clear winner at this time, not even seen as particularly technologically challenging: remember, inverted indices have existed long before Google) and make it a vastly profitable niche, while at the mean time "sucking up" Silicon Valley talent as to create a "moat" around their business. The talent could then yield secondary moats, much like Google's expertise at scaling out on commodity hardware enabled them to create the scalability moat at a time when all other players ran on mainframe hardware (with the notable exception of Inktomi, which started on clusters of bottom-to-mid Sun SPARC hardware, the kind that developers had on their desktops rather than enterprise and mainframe class gear -- quite close to commodity hardware), afraid to tackle the tough distributed systems problems.