How Larry Page Learned Java

The current head of Google, Larry Page, during his student youth was not an outstanding programmer. Stephen Levy tells about the history of creating programs Backrub and PageRank in his book " In the Plex " (quotes: 1 , 2 , there is a link to the first part of the book, which lies in the public domain).



Page could not achieve normal operation of the crawler and indexer due to the large number of bugs in the code and also because of the use of a new and unstable Java language. The system fell all the time and was not suitable for real work. In the archives of the news group comp.lang.java for 1996 preserved questions of Page . He is trying to figure out from knowledgeable people how to assign a User-Agent for the HTTP header.

 Newsgroups: comp.lang.java
 From: p ... @ cs.stanford.edu (Lawrence Page)
 Date: 1996/01/07
 Subject: Q: Setting User-Agent Field?

 I have a web robot which is a java app.  I need to be able to set the
 User-agent field for citizen (so
 people know who is accessing their server).  Anyone have any ideas?

 Right now, Java sends a request that includes something like:

 User-Agent: Java / 1.0beta2

 I'd rather not rewrite all the HTTP stuff myself.  I tried just searching
 in jdk for the java / 1.0beta2 figuring I could just change the string
 but I couldn't find it.  Perhaps it is stored as a unicode string?

 The field user field should probably not be added to
 Java, so people can properly identify their programs.  

 Thanks, Larry Page 

To correct numerous code errors, Page turned to his friend and Stanford graduate student Scott Hassan for help. He was also a good friend of Sergey Brin, whom he met at Ultimate Frisbee student games (a contactless sport in which team players must pass a flying disc to each other).



Scott Hassan says the number of bugs was too big. At first he tried to correct them, but soon he realized that it was a waste of time. So he rewrote the code in Python. Over the next two years, Page and Brin discovered that the new engine could be used on the web and tried to license it to Excite or Altavista, but failed, so they decided to establish their company.

')

Raising their own server, the young company soon faced problems of scalability. Python code processed no more than ten requests per second, so some users did not receive a response from the server at all. The new Google employee, Professor Urs Hölzle (Urs Hölzle), was much more professional in the field of development, so that he once again rewrote the code from scratch, already in C ++.



Of course, the company subsequently hired even more programmers of the highest level, and now the Google search engine is an example of the most scalable and high-performance system in the world, but at first it wasn’t quite like that. To the current state of the company has been for many years, constantly rewriting and improving the system. All this time, the main driver of success was the idea of ​​using PageRank for ranking web pages, the success of which Paget was, apparently, 100% sure. His knowledge of Java was enough to write a working prototype and test the idea, and the rest was already a matter of technology (start-up capital and good programmers).