Variable names should be self documenting
Is Pen Up? Is a question.
Pen Is Up. Is a statement.
Get A Life is one too.
161 publicly visible posts • joined 27 Sep 2009
The managers always complained about the system; it was done by middle level managers (not line, two levels up in the hierarchy) and they were forced to rank everyone, "even if everyone performed exactly the same"; ho ho, aren't managers jolly chappies.
The MS fanout was 3-4, the hierarchy depth was about 7 (well, I was seven levels below BillG, I counted once), for those who can't do math there were a little over 20,000 employees at the time. Using 4 for the fanout there were 16,000 grunts, 4,000 line managers (I was one) and about 1000 middle managers.
From a hierarchy point of view the system worked, and probably still works, because the upper management can only handle a certain amount of information. Whether I or anyone who reported to me was particularly brilliant is particularly uninteresting; what matters is what my manager's reports as a group do. Or, more realistically, my manager's manager's manager, who is stack ranking him (or in a couple of cases only, her).
The downside to the system is that individual excellence or, for that matter, individual stupidity, has no route to the top or the bottom - you live or die along with the rest of the people ("team") you happen to end up with.
I believe Microsoft didn't care - it had a somewhat more enlightened attitude and realized that every one of us can do remarkably good, or remarkably bad, work given the right environment.
Alas somewhere along the time someone fell for someone else's marketing bullpuppy and maybe now believes in the model pursued by a certain other company which I once heard referred to as "prima donna programmer." (That reference comes from around 1994 from someone who never worked for Microsoft or the other company in question.)
John Bowler
No, I think you get a larger one.
I remember, back in 1993, when Nokia had the EO220 - the first smartphone - within the reach of its fingers a bigger beaurocracy; AT&T killed it. Meanwhile, another beaurocracy - the fruit company in question - was killing EO by promising something better tomorrow. (Hands up all those who still use a Newton, ha ha.)
Your whole article comes back to that age old geek question, "why does good software get killed by bad software?" Beaurocracy is irrelevant. Apple has marketing and Microsoft has a company structure that just fixes bugs, bugs, bugs until the stuff actually works. Good engineering with lukewarm marketing cannot defeat good marketing of bad engineering, and simple dumb persistence will win over either. That's why I still keep my Microsoft stock, well, that and 10 isn't a bad P/E today (take a look at ARM Ltd ;-)
John Bowler <jbowler@acm.org>
The analogy doesn't hold on any level:
Microsoft, 1980's: "one one every desktop"
Microsoft, 2004: "ok, done that, what next?" Oops.
Acorn, 1991 or 1992 (internal coffee machine conversation [Hugo]): "one in every washing machine"
ARM Ltd, around 2004: "ok, done that, what next?" "Three in every washing machine?" Ok, that'll work.
As numerous people have pointed out, Acorn morphed into ARM Ltd, possibly inspired by that bit in the Hitchhikers' Guide to the Galaxy where the Marketing People end up on the First Ship. Microsoft hasn't morphed, it's stopped, ceased up.
Meanwhile, for every PC sold one copy of Windows goes with it, and one Intel CPU and, oops, several ARM CPUs and, oops, every flash card and every hard drive has another one...
So, the Microsoft Model of corporate success depends on continuing to develop new and more interesting ways of using the worlds second most complex OS, while Acorn's (now, devoid of the mis-marketing guys, ARM Ltd) depends on someone else finding new and interesting ways of using the worlds most complicated operating systems.
Acorn dissolved because, at least with the marketing strategy it had at the time, it had no future and yet it's market capitalization (shares * share price) was less than the value of the 50% stake it held in ARM Ltd. So it dissolved - Acorn shares became ARM Ltd shares, and Acorn's IP was transferred to Pace.
Likewise Microsoft hit an expansion brick wall - not caused by the total incompetence of its marketing department, as I personally think Acorn's problems were caused by, but because it couldn't expand beyond 100% of the market... so on 2 December 2004 (by my records) Microsoft issued a massive one time dividend to clear a substantial amount of its cash balance.
Unlike other high tech companies both ARM Ltd and Microsoft do currently pay dividends, albeit small ones - about 0.6% for ARM Ltd and about 2% for Microsoft. Unfortunately the analogies pretty much stop there - they are very different companies with very different limitations.
Torben:
>Actually, desktop computers were ARM's original market.
Yes, ostensibly, but Acorn's marketing department had been previously fired by the Sirius Cybernetics Corporation. They (the marketing department) could only conceive of selling into education, and even higher education was an unattainable leap of the collective imagination (hence the fact the UN*X version only sold 200 copies.)
Someone who had been working on the UN*X version uttered a phrase I will never forget; "one in every washing machine."
I actually prefer the even more ironic, "two in every washing machine." The main CPU is an irrelevance - even if every netbook and smartphone in the world uses an Atom as the main CPU there will still be ARM processors, probably *many* ARM processors, in the machine.
The article suggests that Intel has realized this. Back before iAPX86 every microprocessor was an embedded controller - that's why microprocessors were developed (to replace manufacture-time programmed PLAs with software controlled logic). The 6502 was an embedded controller, the ARM was too. Intel has been driving headlong down a blind alley for 25+ very successful years, but it really *is* a blind alley. There is only so much money to be made controlling the main CPU, there is so much more money in all the other CPUs that the main CPU has to call on to do even something as simply as reading a piece of flash memory.