<div dir="ltr"><div class="gmail_default" style="font-family:verdana,sans-serif;font-size:small">Back in the mid-to-late 1980s I was the ringleader of the UNIX underground at IBM Research. Interestingly, we were for a couple of years the largest non-academic customer for Sun Microsystems on the east coast of the US.</div><div class="gmail_default" style="font-family:verdana,sans-serif;font-size:small"><br></div><div class="gmail_default" style="font-family:verdana,sans-serif;font-size:small">When IBM bought ROLM, a maker of telephone equipment, they were confronted with ROLM's insistence on using Sun equipment (and UNIX in general) for their software development.</div><div class="gmail_default" style="font-family:verdana,sans-serif;font-size:small"><br></div><div class="gmail_default" style="font-family:verdana,sans-serif;font-size:small">So a stream of IBM executives made their way to my office in Yorktown Heights to try to figure out whether this demand was for real.</div><div class="gmail_default" style="font-family:verdana,sans-serif;font-size:small"><br></div><div class="gmail_default" style="font-family:verdana,sans-serif;font-size:small">I would show them my development environment (emacs and make plus a bunch of ancillary tools) and demonstrate how I could edit code, build, test, and debug quickly and smoothly.</div><div class="gmail_default" style="font-family:verdana,sans-serif;font-size:small"><br></div><div class="gmail_default" style="font-family:verdana,sans-serif;font-size:small">After half a dozen VPs came through, they agreed and placed a large order with Sun for ROLM. That might have helped the business case for a better AIX, but I'm not sure.</div><div><div dir="ltr" class="gmail_signature" data-smartmail="gmail_signature"><div dir="ltr"><div>=====<br><a href="http://nygeek.net" target="_blank">nygeek.net</a></div><div><a href="https://www.mindthegapdialogs.com/home" target="_blank">mindthegapdialogs.com/home</a><br></div></div></div></div><br></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Wed, Jul 3, 2024 at 7:35 PM G. Branden Robinson <<a href="mailto:g.branden.robinson@gmail.com">g.branden.robinson@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">At 2024-07-03T08:59:11-0600, Marc Rochkind wrote:<br>
> Steve Jenkin suggests: "Developers of Initial Unix arguably were<br>
> 10x-100x more productive than IBM OS/360..."<br>
> <br>
> Indeed, this is part of accepted UNIX lore.<br>
<br>
That claim reminds me of a more general one. Applied to software<br>
development writ large, it seems to be lore, not a reproducible<br>
scientific result.<br>
<br>
I refer of course to Sackman, Erickson, and Grant's 1968 CACM paper<br>
which documented a DARPA experiment that found a productivity range of<br>
28:1 in their sample of programmers (with veterans of 7 years'<br>
experience pitted against "trainees"). Naturally enough, plenty of<br>
people who make claims about variance in programmer productivity are<br>
unaware of this paper's existence; it's not actually relevant to them as<br>
a source of knowledge.<br>
<br>
<a href="https://web.archive.org/web/20120204023412/http://dustyvolumes.com/archives/497" rel="noreferrer" target="_blank">https://web.archive.org/web/20120204023412/http://dustyvolumes.com/archives/497</a><br>
<br>
Thomas Dickey, better known today as the maintainer of ncurses, xterm,<br>
lynx, and mawk (all for 30 years or more, and among other projects),<br>
published a critique of this study in 1981.<br>
<br>
<a href="https://web.archive.org/web/20120204023555/http://dustyvolumes.com/archives/498" rel="noreferrer" target="_blank">https://web.archive.org/web/20120204023555/http://dustyvolumes.com/archives/498</a><br>
<br>
Bill Curtis published a critique of the Sackler et al. paper in 1988.<br>
<br>
I quote (via Dickey):<br>
<br>
"Sackman's ... message that substantial performance differences do exist<br>
among programmers remains valid. Detecting a 20+:1 range ratio depends<br>
upon having one brilliant and one horrid performance in a sample.<br>
However the range ratio is not a particularly stable measure of<br>
performance variability among programmers. The dispersions of such data<br>
as appear in Table I are better represented by such measures as the<br>
standard deviation or semiinterquartile range."<br>
<br>
<a href="https://invisible-island.net/personal/paperstuff.html" rel="noreferrer" target="_blank">https://invisible-island.net/personal/paperstuff.html</a><br>
<br>
We have likely all observed how this 28:1 ratio has bloated in retelling<br>
over time, like the length of a fish catch, to 100:1 or even 1000:1.<br>
Similarly we're all familiar with the common practice of presenting the<br>
mean and sometimes the range of some data sample to support one's<br>
argument, without mentioning the median or mode, let alone the variance<br>
(or the standard deviation). (If a member of one's audience is familiar<br>
with non-Gaussian distributions and inquires whether one's sample may be<br>
better characterized by one, you invite them to disengage from the<br>
discussion.)<br>
<br>
I assert that this "productivity gap" is a myth, and that it persists<br>
because it serves the purposes of diverse audiences who adopt it with<br>
motivated reasoning.<br>
<br>
1. Immature Unix enthusiasts like to reassure themselves, and others<br>
nearby, of their inherent superiority to rival programmers.<br>
<br>
2. Managers like to contrive reasons for (not) promoting individual<br>
contributors. It's easy to cite this productivity "statistic" and<br>
then suggest, without indicating anything concrete, that an employee<br>
is either a rock star or a mediocrity.<br>
<br>
3. Directors in organizations like not having to further justify a<br>
"stack-rank and cut" approach to reducing salary and benefits as a<br>
proportion of operational expenditures.<br>
<br>
<a href="https://en.wikipedia.org/wiki/Vitality_curve" rel="noreferrer" target="_blank">https://en.wikipedia.org/wiki/Vitality_curve</a><br>
<br>
4. Business culture in general is deeply wedded to the idea that<br>
individual productivity, merit, or capacity for "wealth creation" is<br>
variable by several orders of magnitude, because this claim<br>
"justifies" variance in compensation over a similarly large range,<br>
even among college-educated professionals in an organization,<br>
setting aside those members of staff whose collars shade more toward<br>
blue. (Outsourcing is useful in increasing opacity, segregating<br>
workers, and setting them up to have conflicting interests.)<br>
<br>
If people start applying their capacity for critical thought to the<br>
proposition that the CEO is 40,000 times more productive than a<br>
"Software Engineer II", nothing good will happen.<br>
<br>
_Is_ "productivity" among programmers, however defined and measured,<br>
nonuniform? Likely yes. Has our industry studied the question in a<br>
serious way, applying rigorous experimental design and statistical<br>
analysis? Perhaps not.<br>
<br>
And if we did, would any of the people making this claim read or<br>
comprehend the research if it didn't support their biases?<br>
<br>
You already know the answer.<br>
<br>
We utter myths about falsifiable propositions not because we care about<br>
their truth values, but precisely because we don't.<br>
<br>
Regards,<br>
Branden<br>
</blockquote></div>