On Mon, Jul 19, 2021 at 9:41 AM Theodore Y. Ts'o <tytso@mit.edu> wrote:

If anyone can show any examples of people actually *using* the term
"open source" in the sense of "sources are available" before the Open
Source Definition was promulgated, that would be great.  But
otherwise, I think you're trying to retrofit a definition that was
never historically used.

Which is my point -- we never had a name for the behavior, but the behavior certainly existed for years before.  Funny, I just got an email last night from Cerf, Sax and Haverty.

Here it is cut and pasted: 

"Good paper!   As I was reading it, I kept thinking that the same story could be told about TCP, which IMHO  has succeeded for many of the same reasons.

Another possible cause of multiple mechanisms -- the fact that an incompetent novice can make changes to open source.   When I took on the task of writing TCP for 11/40 Unix, I had 1) seen Unix used once and thought the console interactions were pure gibberish; 2) had programmed in assembler on an 11/05 but never did anything in C; 3) had written applications (e.g., email) that used the ARPANET, but had never written any network system code; 4) had never heard of TCP; 5) had done some minor OS work in Multics, CTSS, and ITS, but knew nothing about Unix.   Apparently, those qualifications made me perfect for the assignment.

I suspect there's many similar situations where such people create code and it works its way into the system.

BTW, the multiplicity characteristic is widespread.   I have a handful of machines running Ubuntu, and I'm always amazed at how many different but apparently similar mechanisms exist to do the same thing.   Struggling now with USB, trying to get a new mouse to work the way I want.   Libinput, Evdev, xinput, .... where is Lions' current edition for Ubuntu......."


Which was (in context), a reaction to my observation about UNIX being successful because it was open source and people could use the idea, the code was published, al biet the license to use was not with our remuneration.

This is coming from the networking and Tenex world.   We had the same observation about the PDP-10 and ArpaNET community.  Doug points out SHARE and DECUS.

The fact is anyone that lived in that world will tell you that it really is not that different in behavior or intent.   Yes, DECUS and SHARE had/have a lot of trash -- but you did not have to take it all -- just like today.  Does anyone everything just from the Gnu project much less all the possible apt-get install for Linux?  

Ted -- yes, your generation put a >>name<< to the behavior, which is a wonderful thing and something you can be proud.  But the behavior of openly sharing your work product with the community long predates, Linux, the wider Internet, et al.  It is sad a minimum, if not downright disingenuous to say "open source" was created at that point.

What changed was Moore's law allowed more people to participate because the cost of entry was dramatically lowered.  Remember the cost of deploying UNIX (or Tenex or OS/360 etc..) was completely dominated by the HW cost.   A few $K for an SW license was noise, in large sites a rounding error.

The Internet changed how distribution took place.   Netnews and the like changed how people learn about new things (you did not have to be part of the club).

But in all cases, the same behavior was there and it was just a smaller group of people because the cost of the HW was the barrier to entry.

Clem