Hacker News Reader

HNHacker News

Top New Best Ask Show Jobs

Comment by cperciva | Hacker News Reader

Parent Full thread

cperciva·

>...how about you giving us an example of a non-YC startup you think is doing something really new...

The first example which comes to mind is powerset.

View on HN

Vaporware? That's not really new.

Anyway, they're probably at +3 standard deviations for "newness" compared to all tech startups. Kind of a high bar to match.

Well.. its not so much the vaporware aspect of things... Succeeding in search today is all about scaling. Only the big three have what it takes to build huge data centers. A startup the size of powerset -- not really. Natural language search is something that the average search engine user would consider a "cool feature" at best. The real power of a search engine comes from its ability to cater to the long tail -- which translates to its ability to build huge data centers.

The software involved is more of a key issue. You can always get more machines off the shelf, but you can't just go to Dell.com and place an order for a few killer algorithms. If you have awesome software, you can get funded. IIRC, Powerset is already backed by some big players.

The infrastructure behind Google, Yahoo and even Microsoft is not just a bunch of machines plugged to the same Ethernet switch... They're building datacenters from scratch.

Sure, but it's still an easy problem. Building big data centers is definitely a problem these guys face, but it's by far not the biggest or hardest one. The software infrastructure and algorithms are much more daunting problems.

Let me remind you that Google was a small startup very recently, and that was during the "dot com bust".

Natural language search is something that the average search engine user would consider a "cool feature" at best.

But natural language search that works would be HUGE.

Reply on news.ycombinator.com