[TUHS] Why Linux not another PC/UNIX [was Mach for i386 ...]

[Dan Cross]

[Apologies in advance for this enormous wall of text]

On Wed, Feb 22, 2017 at 10:36 AM, Clem Cole <clemc at ccc.com> wrote:

> Dan & Larry thank you -- this helps me understand and I'm going reply you
> both in line hopefully without screwing up either of your messages as I
> try...
>
> On Tue, Feb 21, 2017 at 11:28 PM, Dan Cross <crossd at gmail.com> wrote:
>
>>
>> The universities you are mentioning here are top-tier for CS. But please
>> do bear in mind that if you were not at one of those institutions (for
>> whatever reason), asking for that code might well have gotten you the hairy
>> eyeball from folks you didn't want giving you a furry look.
>>
> ​Unfortunately, I can see that.   Sad, but probably a reality.​  Again,
> "he who has the gold, rules."  Funny thing about gatekeepers.  Larry's
> closing comment about Bill Shannon walling off the Research kernels was the
> same thing, and IT folks often seem to be like that too.   My wife likes
> referred this behavior just this AM at breakfast, she calls "can't" a
> "magic button word" for me.  I hate it when providers say things like
> that.  Pisses me off and something go off to prove otherwise. ;-)
>

Well, for what it's worth, I think that all parties were acting in good
faith. Sure, there was a little bit of a "ooo and ahh" factor to have root
access or access to source code on what were, at the time, expensive
machines, and that lead to something of a "cabal" forming that felt like
they needed to protect such things.

Still, I think the bulk of the hesitation had its roots in three things:

1) Protecting university resources. Hardware often came from research
grants, and the bulk of the school's research wasn't in CS. The MechE or EE
departments, for example, didn't really want Unix hackers using machines
that had been obtained through a grant to study beam stresses or molecular
properties of semiconductor materials. Could you imagine some poor grad
student 16 hours into an 18 hour simulation run when his/her professor's
workstation rebooted because some joker thought he had a better terminal
driver or something goofy like that that just absolutely had to go into the
kernel right then? (I mention that example specifically because Ken Arnold
has a great vignette on his web site about hacking the TTY driver at UCB.)

Similarly, access to wide area networks was mainly to support research, not
give budding CS weenies FTP access. I mean, I kind of get this: being
irresponsible with a grant could get you in trouble with the funding
agency; similarly, starving out bandwidth for your researchers so that they
had trouble communicating with their collaborators at other institutions or
with their sponsors would give the administration headaches as they got an
earful from a large community of angry professors and grad students. And at
the time, for geographical reasons, getting more bandwidth to State College
would have been painful.

Example: I found a discarded VAX 750 in a room in the physics department
one day; someone had put 4.3BSD on it. I fired it up and a physicist nearly
had a heart attack, "You mean that thing is running?! OMG; shut it off
before someone sees you!" I'm pretty sure they were thinking that this was
not exactly in line with the department's core academic mission.

2) Protecting the university from litigation. Let's be frank: students
(and, uh, young townies who hang around too) could be careless and don't
always exercise the best judgement (*cough* not that I would be talking
about myself or anything *cough*). I could imagine that sys admins who were
not lawyers might be paranoid that they could endanger their jobs and their
ability to support themselves and their families by opening the university
to some legal danger by giving someone access to something they shouldn't
have access to (e.g., Unix source code). If I thought my job was at stack,
I'd be hesitant too.

3) Protecting themselves from being overburdened supporting someone in
waaaay over his/her head. I managed to wrangle a tape of the SPARC port of
4.4BSD (encumbered) out of someone and get it installed on a SPARCstation;
I was a high school senior. I had to promise to a) not give it to anyone
else, and b) leave the person who gave me the tape the hell alone when it
didn't work (though it did). In other words, he'd give me an 8mm exabyte
tape if I never mentioned it to him ever again, but only because I knew
him. I thought that was a fair deal.

There also just wasn't that much of a Unix hacker culture there. There were
a few folks who were exceptionally good, but they'd graduate or otherwise
migrate away. And by the time I came on the scene, we were in that odd
phase where the most advanced systems were commercial and vendor supplied.
I remember suggesting that we put BSD on a bunch of the Suns in one of the
labs, and getting shot down because it wouldn't be supported by Sun. I
found that odd since only a decade before the same group of people didn't
seem to bat an eye at getting a tape of 4.2 or 4.3 from Berkeley and
putting that on a VAX that cost something like 10-20 times what a
SPARCstation cost and running the whole department off of that.

For myself, early on, I may have encountered resistance more because I
wasn't actually a student (though I was nominally "staff" in the sense of
being a university employee, albeit minimum wage...hey, for a high
schooler, it sure beat what my buddies who were washing dishes or flipping
burgers were doing to earn gas money). I ended up going elsewhere for
college and majoring in math instead of CS, but by then it was clear that
Linux had won and it wasn't an issue anymore anyway. I was running Plan 9
at the time, though.

[snip]
>>
> As someone once said, BSD is what you get when Unix folks port to the PC;
>> Linux is what you get when PC folks build a Unix.
>>
> I love it, never heard that and in fact that helps with Noel's original
> question, I think.   It all comes back to the Christiansen disruption
> theory.
>

I think the disruption was also cultural and not just technical. One of the
most maddening things to me about the Linux community is that they don't
seem to listen to other folks or critically examine prior art. I know
that's a massive generalization and it's obviously not true in all cases,
but I think it's fair to a first order approximation that they often "go
with their gut."

I think at least part of that is because, when Linux was getting started,
the existing Unix community was somewhat haughty and aloof and sort of
looked down on it as a toy that would never go anywhere. As a result, I
sort of feel like they've got something of a chip on their shoulder about
other ways of doing things; the prevailing attitude seems to be one of,
"well, it's right because that's how we do it." Epoll is a great example of
this; it has all sorts of strange structural problems that didn't exist in
BSD's kqueue, but adopting kqueue was unpopular. Perhaps that stems from
being told too many times that what they wanted to do was alternately
impossible and stupid. They've proven the "other side" (for whatever that
means) wrong enough times that when they're told there could be another way
it's immediately dismissed. And while skepticism in general is probably
good (the 'S' in 'CS' is for 'Science', after all, and a healthy dose of
skepticism is important to look at something scientifically...), I feel
that the pendulum has swung so far it's now just incredulity that someone
would dare contradict the way Linux does something.

​...
>>  A self-deprecating anecdote.
>>
> ​I had to laugh a little when I read all that.   I'm going to reply to
> something Larry said in a minute and this all related.   Yeah, Larry's
> right, places like CMU, MIT, UCB are elite schools and yes, I have too
> solid board scores *etc*.  As I like to say I have "the usual degrees
> from the usual institutions" - *i.e*. I have my union card.  But I'm
> nothing special.  You're from Penn State or UWisc (aka "UC Madison"  - a
> lot of my class from UCB is the core of the faculty there).  Hey,  I
> believe Seymour Cray did his undergrad at St. Olaf's, a school better know
> for music - i.e. a small liberal arts school in Northfield, MN.
>
> I've never really cared where you went to school, what your score were,
> what your degrees are etc.  I'm a hacker, and proud of being that.   The
> schools, as you and Larry correctly point out, gave me opportunity and
> access.  So I have network from them.  But its what you do with it that
> matters to me.
>
> Two stories about me.   First, I have always said, the greatest gift I was
> ever given was *not* getting a scholarship to MIT.  I would have gone
> there and likely been "The nerd down the hall" - either that or flunked
> out.  Who knows, as I later got to know folks that went there, it would not
> have been a good match for me as an undergrad.  CMU (as screwed up as it
> was at time) was a better match for my personality.
>
> The fact is, I did not know know enough about the MIT culture when I was
> in HS (I was a faculty brat - *i.e.* scholarship student -- from a
> Philadelphia prep school - my Sr year in college 7 of the 7 Ivy League
> Squash Captains are my classmates from said prep school).  That HS pushed
> me to MIT because I wanted to an engineer and that's all they knew.  I did
> not even know about CMU until it was suggested by a family friend who was
> professor in the B School there.   But in the end, it was about $.   CMU
> offered me a scholarship, MIT did not and tricky Dick wanted to put a gun
> in hand.  It was an easy choice.   What was lucky for me was it a
> reasonably good culture match...  mostly because of the close friends I
> made there ...  out side of the EE, Math and CS Depts (two weekend ago I
> was a party with some of them that has occurred for 40 years on the same
> weekend since).  Point is, I got lucky...
>
> Second, the proudest moment for me was watching my children pick
> colleges.  Unlike me, I swore they would know about the culture of the
> schools and make darned sure that where they went matched their
> personalities and not rely on luck (and I'm very pleased to say that worked
> well with my daughter and seems to be working with my son).  So to me, what
> the school you one too says about you is the network you have, who are your
> friends and the culture you learned.  It tells me a little about how I can
> expect you to have been versed as a starting point, but I'm really much
> more interest want you do, have done.
>

Great stories. I understand where you are coming from. The culture thing is
critical; that's why I decided not to go to Penn State; having lived in the
town I knew it wasn't a great fit for me. Also, to be frank, I wasn't
really ready to go immediately after HS. A couple of startups, a stint in
the Marines and experiencing M.O.Rabin teach cryptography at Columbia (he
was on sabbatical from Harvard) eventually got me walking in the right
direction though. :-)

It's sad, that Penn State and UWisc had walled areas like both
described.  Sadly
> I saw the same thing at UCB, certainly of the undergrads.   I have nothing
> but respect for the young folks that did an undergrad at UCB, because it
> was definitely different as a grad student.
>

To be fair, I think a lot of it was that those schools have a different
focus. The educational focus is on grad students and the academic focus is
primarily on research. I don't think that's bad, but the educational
mission towards undergrads is just different: the state schools do an
excellent job of taking in-state kids from farms and formerly-factory towns
and giving them a solid education.

A colleague in my office now dropped out of a math PhD program $n$ years in
ABD and made a fascinating remark to me once; he asserted that the way to
REALLY learn math was over coffee with a professor or with fellow grad
students, not sitting in a classroom or studying texts or the literature. I
remembered that from my own perspective, some of the most eye opening
experiences accompanied by the largest number of "Aha!" moments *I* had
were sitting in the common room chatting with my advisor about various math
topics, or taking something simple I'd done to a professor and asking for
commentary. But certainly not sitting in the lecture hall in Mathematics
building furiously transcribing what the learned professor had written on
the board into my notebook. Well, when you've got a huge undergrad
population logistically that's just not feasible for everyone.

  To me that's about respect for the individual and helping them grown up
> to be their best - creating opportunity.  But I fear you are right.   If
> things like UNIX access were walled off at places like that, then as you
> both point out, people we search for it where they could find it, *what
> is sad is that BSD UNIX was available at the time Linux was available.  *The
> problem was that too few knew it, although many did  (more in a minute).
>

Yup. It really feels like a lost opportunity.

[snip]
> I think your counter point is that while I believe folks like yourselves
> or Linus could have gotten BSD UNIX if you had tried to find it it was
> available and folks like Keith and Bill were trying to get it out the door,
> but  have suggest that you don't think so.   You think the walls were too
> high, the access was only for the "chosen few" and a difference was the
> Linux really was available to what Larry referred to as the great unwashed.
>

The unwashed masses comment is apropos. A few places I've read papers
advocating that one should look at Unix the way one would go about literary
criticism; in this sense, Linux feels more like a populist reaction to Unix
than a linear continuation of the same school of thought. Indeed,
comparisons to the arts abound in our community. Indeed, Stu Feldman wrote
a neat metaphorical paper comparing the history of Unix to the history of
western architecture (
https://books.google.com/books?id=5xyk_PXaloAC&lpg=PA8&ots=cZQ3TYbp71&dq=stuart%20feldman%20architectural%20history&pg=PA8#v=onepage&q&f=false
).

        - Dan C.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20170223/16ff0f74/attachment.html>