On when to read the scientific literature [?]

from an interview with Irving Weissman:

Probably the best advice came from the late Donald Michie; he was both an immunologist and one of the founders of the field of artificial intelligence. He told me in 1962 that when he began a project, he quit reading the literature. He'd work at his project until it was done, and then he'd read the literature so he could write up what he was doing. What that meant to me was that the greatest pleasure is being the scientist who is doing the science, not looking over your shoulder to see if you can make it go faster.

UPDATE. One IPBiz reader wrote in:

I'm appalled that someone could say

Probably the best advice came from the late Donald Michie; he was both an immunologist and one of the founders of the field of artificial intelligence. He told me in 1962 that when he began a project, he quit reading the literature. He'd work at his project until it was done, and then he'd read the literature so he could write up what he was doing.

Now, obviously someone had been reading BEFORE he began a project..but to keep working without reading the literature means
* duplication of efforts (a waste of $$)
* a poor example for any grad student
* he could actually fail to recognize his own errors and go down wrong paths.

I did know one Berkeley prof who wrote and won a proposal, without reading the literature. After he won the proposal, he looked at literature, and found everything he had proposed to do, had already been done!!

IPBiz notes that in the world of patent law, infringement is strict liability; one cannot say I'm innocent because I didn't know about the patent. This gives an incentive to read patents, notwithstanding what Torvalds has said, Reading patents, and the scientific literature, helps avoid duplication of efforts,