Comment by dekhn
2 days ago
Yes, but the downside is even highly accomplished engineers feel unworthy. For example, when I started in 2008, they said everybody should make a "Google Resume" (ongoing list of all the stuff you did at Google) and linked to Jeff's as an example.
He rewrote the entire indexing pipeline at a critical time enabling Google's rapid growth... created mapreduce... helped create bigtable & gfs... wrote the search engine that ran for over a decade... numerous improvements to search and ads quality (back in the days when search and ads quality meant something)... and that was just the first few years.
Everyone at Google has imposter syndrome. If you don't, this is how they make sure you get it.
Wasn't Peter Norvig one of the key people behind Google Search? Would it be correct to say that scientists did the research, maybe reference implementations in Lisp or Python, and later Jeff Dean did the actual C++ production?
Peter worked on search, yes. I think he mainly focused on quality. See his comment on "Managing Gigabytes" (https://www.amazon.com/review/R33MIMY7A7C2H8/ref=cm_cr_srp_d...) which was an early reference for the compression techniques used to make search fast.
The way jeff described it was that there was a very early contingent who took Larry's original code and made a production system out of it, but it wasn't very good- the indexer basically had to be run all the way through without errors to make a new index- and that was delaying a release of a fresh new index, so results were getting stale. At that point Jeff "invented" mapreduce (the shuffle stage was Sanjay) and wrote the new indexer and the search engine as well.
I'm not aware of any LISP implementation of the Google search engine; if there was one, it would have been extremely early, like in the stanford days, I think.