>> IBM went through a relatively brief period of stagnation. It did get
>> too bloated by hiring too many people to fit the IBM "no layoff" model
>> and ceased being lean.
> It was not relatively brief. It was continuous. But it wasn't
> stagnation as much as taking great things and doing mediocre things
> with them. It was just that being dominant in their fields hid much of
> it most of the time. I have friends in some key places inside IBM and
> they talk about the constant battle to compete against the
> competition, not against what they did last year. But it has gotten
> much better. Read the Mythical Man Month. He talks about how only IBM
> had the resources to do the System 360 at the time but the bureaucracy
> of IBM almost doomed the project.
I read Mythical Man Month some time ago but don't recall that the IBM
bureacracy almost killing the project. IBM was a much smaller company
Could you elaborate on what was said on that subject?
My impression from the book (and others) was that IBM greatly
understimated the time to develop S/360 hardware and software, test
it, and get it out the door. It was hard to test the software before
the hardware was ready to have something run on it.
Further, the OS software turned out to be much more complex than
expected, that is, harder to write, harder to debug, harder to use,
and most importantly, extremely slow. Brooks' point was that throwing
more people at the project only made it later, and some projects will
require time so matter how many people are thrown at it.
> When Gerstner came in ....
I think the biggest problem when he came it was IBM was too big to
support the declining business. If there were no PCs or minis and IBM
kept selling bigger and fancier mainframes ever year -- as it had been
doing since 1952 -- it would've done ok. But that was no longer the
marketplace. Just like GM discovernig the marketplace of the 1950s
and 1960s had expired in the 1970s.
Mainframes were more of a commodity than they were before; IBM's
extensive customization options were not as necessary. I remember the
extremely complex option list for PCs -- stuff that would make sense
on a mainframe but not on a PC.
In other words, while mainframes were still a viable product, IBM had
to significantly reduce the number of people involved in making
them--reduce the cost per unit.
IBM never had to contract before in its history, it never had layoffs.
It almost always had growth.
The consent decree (more below) constrained IBM from freely competing
in those years.
> But try going into an IBM retail store in the 80s and buy something
> for PC. Before you could pay for it and leave you had to register
> the serial number of the floppy disk drive with your name and address.
A lot of software vendors in those days imposed various restrictions.
> At to Bell Labs, they did great work. Most of it took years to make it
> into the field. Much of it went no where in terms of the phone
> system. I still remember how colored phones were marketed as
> innovations. I had an older cousin who was a senior sales rep for Bell
> in IL in the 70s. He could spout the company line all day long about
> innovation, protecting network from foreign devices, etc ... He would
> also say privately it was all about locking everyone else out of the
In the 1970s, with electromechanical switching, there were very
legitimate concerns about foreign devices hurting the network. The
Independent phone companies, state PUCs, and FCC engineers were all in
agreement on that risk. ESS has more tolerance and controls.
Further, the world has changed. In those days, regardless of actual
fault, Bell would be the one held responsible -- blamed -- for any
problems caused by foreign gear, not the actual irresponsible party.
As to keeping others out of the market, that was FCC policy. The
market was cross subsidized -- the profits from business and toll
services cross subsidized local residents. (Today we have 911 and
Universal Service Fees. Back then the dial 0 operator handled 911
calls and that was a big part of their traffic.) Anyway, they knew
that they'd STILL have to keep residential rates low but without the
> It's called a free market. If they really impinged on IBM patents they
> could have been made to pay royalties or not sell the products.
IBM, as a result of the consent decree, did not have patent
protection. It was required to release its patents and technology to
allow that kind of competition. But how far should IBM go to prop up
> What really was happening was IBM was claiming that plugging in any
> non-IBM device voided the warranties and service contracts on the
> entire system; similar to Bell and their $150 + monthly charges data
> protection device to plug in a phone not supplied by the phone
> company. Or heaven forbid a modem. If the old Bell system was still
> in place without the fall out of the MCI situation, we'd all still
> be debating whether or not 128kbps ISDN was worth the extra costs
> and wondering why the rest of the planet had 3mbps Internet. But
> we'd be told over and over it wasn't very reliable compared to the
> great product we were using. Heard that line WAY too many times.
Times have changed since 1983. IBM changed over the years and the
Bell System would've changed to. It was already on its way to dumping
company ownership of customer equipment with ESS coming in, better FCC
certification, and a recognition that customer problems were the
customer's fault. IBM prospered when it dropped bundled and so
would've Bell -- as the Baby Bell's did quite nicely.
One key difference was that IBM won its lawsuit but was content to
stick with the consent decree a while longer (it eventually rid itself
of that and made a big difference). AT&T want it gone so it could go
into new projects, which it promptly did.
> Watching the mainframe sort wars of the 80s was similar. Syncsort(?)
> had a sort that was up to twice as fast as IBMs. Every week in
> ComppurterWorld they'd run a different ad showing how their sort
> ground IBM into the dust in a different situation. After a few years
> IBM started running ads showing how they really were better
> according to their "fair" tests.
Strange. I remember Syncsorts ads very well; but never saw an IBM ad
for Sort. I also never knew a site that didn't use Syncsort. I
thought IBM's sort was free, anyway. Software products didn't become
a big thing until the 1990s; they were once free, then pretty cheap.
> Bell and IBM had great research arms. They had lots of bright people.
> But the companies were run by managers who were out to protect their
> turf at all costs. And bringing new widgets to markets where they
> already owned 90% of the market was riskier than doing nothing.
Except both companies continually did bring new products to market.
Also both companies made continual improvements internally to existing
products. The "bank per buck" of IBM machines always improved over
> I say the above based on a variety of information sources. I've been
> a potential customer of IBM where they wanted us to switch to them
> in the 80s for the 500 or so minicomputers we were installing around
> the country a year.
I do not claim that every IBM product was the best in the marketplace
or IBM didn't make mistakes. However, my employers have been IBM
customers for the last 35 years and I've seen what they can do. I've
also dealt with other vendors and for the most part IBM was superior.
Not always, but usually.
> Agreed. But then again IBM and AT&T were never in a hurry to put out
> new things unless it would increase profits.
Just like any other company.
> And new things in technology tend to lower profits unless you also
> grow the market and that's not always a given. If you want to really
> see the economics of high tech at work, look around and think about
> the turn over in retail computer stores for the last 15 years. They
> just have a real hard time dealing with the costs of their products
> dropping by 1/3 year after year after year. It makes you scared to
> put out a new product. And if you "own" the market you tend to do
> just that. Sit on things that are disruptive.
If that were true, we'd still have SxS or panel and open lines and talk
on candlesticks. Period. We'd be using punch card machines. Period.
Both IBM and Bell took plenty of risks over the years. ESS and S/360
were HUGE expensive risks, fraught with tons of problems along the
way. They have taken big risks and small risks.
> As to a counter to this think of Intel, Cisco (usually), Texas
> Instruments, and lately Motorola.