Shannon’s post set me thinking about the odd & perhaps correct clock maker. And it took me back to 1983. We decided to computerize our typing service; my sister visited with the salesman (she ran the business while I had my middle child). As in so much, I think she made the correct choice: we both liked the TI models better but went with IBM, which appeared more flexible and accessible. We needed equipment that several part-time typists a day would work on, typists who came and went for a semester or two.
First, I’m curious about our impression: was it right? Second, if this was true, was it an example of technology that might have been better but because it was less compatible and flexible, did not succeed? In other words, the best clock might be difficult to fix because it is unusual. Of course, I’m sure that an ego can get in the way of common sense. (My brother was convinced of the importance of an invention – not his own so I doubt his ego was heavily involved – for which his company owned the patent; he wanted to buy them out & manufacture it. They didn’t want to sell it nor use it.) Good products for various reasons that have little to do with their quality can fail to find a market (cost, for instance, or a really lousy sales plan).
A month or so ago, we watched Tucker: The Man and His Dream; surely conspiracies by the competition are not common, but, then, that, too, is another question.
I think your choice to go with IBM was the correct one for a variety of reasons.
I had been working with mainframes for about a decade when I got my job with the police as a fingerprint tech. The mainframe used by ID was a Burroughs, and I had to learn everything all over again. I could figure out how to use other computers that had been influenced by IBM’s interface design, but that Burroughs beast was like something from another world.
Are you a Heinlein fan? I found it rather amusing that he will have mention that the sentient computers which fly spaceships in the future are made by Burroughs, a firm that disappeared in 1986. (It merged with Sperry to form UniSys, another mainframe company.)
I think you were very wise to go with IBM. Those part-time typists wouldn’t have gotten much work done, and their error rate would have been high, while they tried to climb up the learning curve on TI machines.
James
I bought a TI product when I was in college. It was marginally affordable with the rebate offered. TI screwed me on the rebate. I never bought another thing from Texas Instruments.
For word processing in 1983, I’m surprised that Wang Laboratories wasn’t in contention. Although IBM dominated the mainframe business, it was late to the party in WP, and Wang grabbed a lot of share.
I just re-read “Riding the Runaway Horse,” by Charles Kenney, which is a well-written history of Wang’s creation, growth, and decline. It’s the old story of a company successful at one level of a technology which fails at the next level, aggravated in this case by a founder/CEO who was committed to putting his son in a senior position.
The question of why a particular technology or form of technology succeeds in the market place is a very interesting one. I would say that, as a rule of thumb, the most successful product is not the best of the best in some particular attribute but rather the one that is consistently second string across several attributes. If we think of grading different aspects of a product on a grade school A->F scale, the product that pulls in solid B’s or even C’s across the board tends to succeed whereas a product of mixed A’s and F’s fails. An all around good enough product beats a mostly perfect product.
I think a balance of trade offs eventually determines success. An old engineering jokes says that any project can have attributes of being cheap, fast and right. Pick any two attributes. In a similar fashion, cars can be cheap, high performance or stylish, pick any two. You can build your own hotrod that runs like hell, cost next to nothing and looks like crap, You can buy a cheap, stylish little toy car that can barely get to speed on the freeway. You can mortgage your house to buy high performance, stylish Italian sports car. The most successful car models will actually balance all three attributes. Moderate price, moderate performance and moderate style beats out all other combinations. In its time, the model T wasn’t the cheapest car, the best performing car and it definitely wasn’t he most stylish yet it traded off all three attributes to create a superior product.
I think we see the same effect in computers. Computers might have attributes of cost, ease of use and software availability. Moderate cost, moderate ease of use and moderate software availability beats out other combinations.
Shannon…one factor that can help determine marketplace success or lack thereof is the *relationship* among multiple products in a product line. In the IBM case, the competing mainframe companies (including Burroughs, Univac, NCR, Honeywell, and GE) had a decent chance *until* IBM introduced System/360, offering a product line which was at least theoretically compatible across all models. This radically changed the terms of the competition. It was a gutsy and high-risk move, and came pretty close to failure. Good discussion in Tom Watson Jr’s excellent book “Father, Son, & Company.”
Another factor, which generally gets far too little attention from strategists, is the quality and configuration of the sales force. A sales force may be excellent at selling one product line, but fail completely at selling another line which on the surface seems quite similar. Yet the economic and political arguments for using an existing sales force, as opposed to creating a new one, are often hard to overcome.
I’ll bet that the keyboards for those IBM’s are still in use somewhere. You would have had to have replaced the TI stuff by mid 86 at the latest.
David Foster,
I agree with you point about the network effect and sales. However, even in those cases I think you will find that an company that reaches overall “good” levels of performance in all facets of its business will beat out company that is “great” in some facets but “sucks” in others.
Shannon, what you say makes perfect sense. The comment from “Fred” illustrates the concept nicely.
Here’s a cool link for those interested:
http://www.nist.gov/
And here’s a sort of urban myth story that’s been making the rounds at NIST, BIPM, USNO, PTTI, and elsewhere for the last decade or so.
There are all sorts of working groups of scientists and researchers from all over the world that get together to improve the basics related to their specialised fields. These various working groups, depending on the military value of their produce, are managed by either institutional managers, or by military personal tasked to oversee the chaos that is 1000 post-docs from all over the world. And one of those military grade groups was related to “time” both scientific and computational, overseen by the USNO back a decade ago.
The myth goes something like this. (it’s kind of weird, so take it with a grain of salt). The Navy Lt. at the USNO in CO. was notified by his sysadmin that there was someone who was accessing a fairly low grade of classified datasets and algorithms in the bins on a server that was exclusively used by scientists, physicists, &etc. working on a very esoteric area, and this hacker was busy downloading ONLY the uncorrected, raw datasets (and correction alogrithms) while ignoring bins that any expert would know were the important ones. So the Navy personel, curious, did their jobs and let this obviously crazy hacker do what he was doing while they had their security people trace the idiot.
Then something curious happened. As soon as the idiot had finished downloading these bins, the Navy Lt.’s email inbox dinged. It was an email from this exact hacker innocently requesting to be included in the professional working group, which of course would allow this hacker to legally have access to the exact same bins that he’s just stolen. This amused the Navy and facilities people very much, as the hacker’s email told them what their security people had already told them several hours earlier (that he was just a hacker and a thief), and provided his home address and phone number and etc. All without mentioning that this hacker had ALREADY stolen what he was so politely and formally requesting. So the Naval Officer ignored his security peoples dire predictions and saved himself a LOT of paperwork by simply putting the hacker on the working group as if the whole thing never happened.
So no physicists or researchers or etc. knew when dealing with this hacker the details. They just assumed he was a post-doc physicist or engineer like themselves, perhaps a bit odd, as the Navy and hacker were very mystereous when it came to crediting contribution. Mostly the guy just asked “stupid questions” and pestered old retired Bell Labs guys, who in turn just referred this fellow over to MIT or the U of Michigan or etc. to whomever was working on some different current group. As the physicists and researchers often didn’t feel comfortable publishing without thanking a contributor, some even made up names for him (kind of inside jokes)… he never showed up to conferences, or used his “official” contact information… How does one deal with a “good will hunting” situation where in real life the “janitor” who wrote down the correct answer is never found?
People, being superstitious as physicists are proud, often just wrote their experience off as a fluke, or told a simple lie to explain what happened because telling the truth was too much of a headache. The MIT professor tells the class it was all a joke. Others give the credit to a collegue that’s just died, a sort of last tribute.
Personally, I think it’s a bunch of DARPA physicists idea of a great joke. As they’re constrained by the agreements they sign with the military, they adopted a funny story about an idiot hacker they were told by a visiting officer and use it as a way to entertain themselves when they’ve hit a wall on their own research. I asked one of those guys once if this was the case, and he was very coy, which annoyed me. So I told him the following to mess with him:
In terms of complex scientific base unit definitions, it is rarely usefull to use a mathematical formula because there are so many unknown variables that a neat equation doesn’t leave room for refinement or isolation and inclusion of those variables. It’s just not very flexable. An Algorithm, on the other hand, when used to define a scientific base unit, affords one the best of mathematics (in the form of function calls) AND allows for the insertion of new discoveries and etc. without requiring any theoretical gymnastics to explain the improvement. Moreover, the use of algoritms rather than numerical equations provides experimentors with a step-wise description of PROCESS that parallels the process of the actual experimental apparatus currently used… so rather than being stuck with a silly equation dating back to the 19th century, an algorithm can merely mirror the state of the art, technologically speaking. Sure, this is a pain for theorists who’d rather be meta-physicians than deal with the actual material universe, but then… so what? Those theorists entertain the undergraduates and say what one will about snake oil salesmen, the placebo effect has its charms.
This is why the use of an algorithm to define the scientific base units has worked so well. It doesn’t replace numerics outside of the realm of theory, and is much more handy when experimentalists and engineers ACTUALLY need to know how the real world operates to improve whatever system they’re working on.
I suspect the DARPA guy didn’t realize I was just messing with his head, and that I was repeating something I myself had read on the internet. Because I got a thanks mention on a paper his group published that I don’t remotely deserve. Oh well. For the record, I’d just as soon he credit the elves who are really responsible…. ;)