by Ken Kolence (as told to Luanne Johnson)
Copyright ©1998, Luanne Johnson, All Rights Reserved
Boole & Babbage was founded in September, 1967. It was the successor to K & K
Associates which Dave Katch and I started in May of that year. We had both previously
worked for Control Data and were pioneers in developing engineering approaches to software
development, which was a new concept at the time. As a matter of fact, I coined the term
"software engineering" to describe what we were doing although I later learned
that the term was also being used elsewhere. To the best of my knowledge, though, this was
the first use of the term in Silicon Valley.
I had also specialized in developing software that would measure computer performance
and we thought that there would be a big market for that kind of software product in the
rapidly-growing IBM 360 customer base. So the plan was to sell our services as software
engineering consultants and get the money together to be able to develop some performance
measurement products for the IBM 360 market.
It didn’t take us long to figure out that we were going to need substantial funds
to be able to develop the software products so we began looking for a venture capitalist
to back us. Software product companies were a really unknown concept so we had a hard time
finding a venture capitalist who would listen to us. But through some contacts I had, we
met Franklin Pitcher (Pitch) Johnson who was active in Silicon Valley investments. He
liked our concept, put together a group of investors, told us to come up with a new name,
and Boole & Babbage was born.
The first product we released was PPE, the Problem Program Evaluator. It was designed
to analyze the performance of a computer program and detect ways in which the code could
be changed to improve the run time, that is, make the program run more efficiently.
PPE was a big success. Our marketing strategy was to say, OK, sign a contract and
we’ll come out and run PPE on any three programs you give us. Then within 24 hours
we’ll give you the corrections you need to make and rerun PPE and if we haven’t
saved you at least 20% of the run time, you can keep the changes we recommended and
there’s no obligation. Well, we sold the program everytime we did the demo because,
in most cases, the run time improvement was anywhere from 30% to 50%.
Now, the last thing we wanted was to have a run-in with IBM, so we deliberately
crippled PPE so that it wouldn’t process compilers or IBM system software or
utilities. It was designed to process the programs that had been written in-house by the
customer’s staff and wouldn’t work if they tried to use it to analyze the
efficiency of the IBM software.
It was our second product, CUE, Configuration Utilization Evaluator, that got IBM
pissed at us. CUE was designed to analyze the performance of the hardware, including all
the peripherals and the channels between the CPU and the various I/O devices. And CUE gave
the IBM account reps major heartburn.
In those days, IBM had various levels of account control. Level 1 meant that the IBM
account representative, the salesman, did all of the equipment planning for the
customer. Literally. Customers were spending a lot of money on computer systems and the
performance was horrible. So the salesmen were selling faster and faster CPUs, which were
expensive, and not upgrading the I/O, which was a lot cheaper. And with CUE you instantly
found out where the bottlenecks were, which often meant that the system performance could
be greatly improved just by upgrading the I/O without replacing the CPU.
Well, after this had been going on awhile, IBM invited us to do a demo of CUE for them,
which we rather naively did. They were impressed with how effectively the program could
identify constraints on performance. So impressed, in fact, that they assigned several of
their field engineers to write programs which would do the same thing. And then they put
their programs into COSMIC, the NASA system for redistributing program code, and told
their customers that they could get performance measurement software from COSMIC for only
a couple of hundred dollars, compared to the $5,000 we were charging.
We protested to IBM and to the SLAC (Stanford Linear Accelerator Center), which was
responsible for maintaining COSMIC, but it didn’t do any good. This was when software
was routinely given away free by hardware manufacturers and commonly exchanged between
computer users at prices that just covered the cost of reproducing and shipping the source
code. Freely distributing software written by IBM field engineers was accepted practice
and they didn’t see any reason why they shouldn’t proceed to do that with a
product that competed with CUE.
This was a very deliberate decision on IBM’s part. Within the same month,
virtually every IBM account rep went to his customers and told them of the availability of
the product through COSMIC. Naturally, it killed the market for CUE because we
couldn’t compete at the price charged by COSMIC which only covered their distribution
costs since the development costs were absorbed by IBM.
Boole & Babbage survived that setback and is still thriving today. As a matter of
fact, it just celebrated its thirtieth anniversary and is one of the very few software
companies started in the 1960s that is still surviving today (ed. note:
Boole & Babbage was acquired by BMC Software in November, 1998). But I think
the story of how IBM killed the market for CUE illustrates very well what fledging
software companies were up against in the early days of the software industry.
 top of page
 return
to the list of anecdotes
|
