![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
Appeared in Volume 6/1, February 1993
M.H. van Emden
LP is a powerful method for making computers do what we want. As computers can do useful things, it would seem that LP has great commercial potential. Yet it remains a largely academic pursuit: to prove theorems about it and to publish papers on it. To see how LP can fulfill its potential, we should consider it from the point of view of marketing.
Marketing must invent complete products and drive them to commanding positions in defensible market segments.
There are several key concepts in this brief statement: market segment and product. I will talk about products first.
A key notion of marketing is the distinction between a device and a product. For example, a transistor is a device, which of course had to be invented. In the typical American style of the eighties, if someone were to invent the transistor now, he would immediately leave his employer, rush off to find a friendly venture capitalist (and a lawyer to fight off his former employer), and set up his own firm. This company might make a decent living selling kits containing a Transistor and some documentation. The market would consist of electronics hobbyists keen to experiment with the amazing new device.
Wisdom borne of hindsight tells us that such a firm would only be exploiting a small fraction of the potential of solid-state electronics. We now see that the greatest impact of the device is via products. Initially these included hearing aids and pocket radios, where the distinguishing features of the new device made the biggest difference. Each of these products was an invention in its own right. It was a complete product: ready for use by the customer, who did not have to know anything about electronics and did not even need to be trained to use it.
It is easy to recognize a hearing aid as an invention. A pocket radio is not as obviously one, but is an invention nonetheless. Perhaps you are only willing to recognize the radio as an invention, but dismiss a pocket radio as mere re-packaging. Well, some of the most important inventions are new forms of packaging. These lead to new concepts in users' minds, and hence to new behaviours. In this way a product can change the world. To let you get used to this concept, let me present another example.
Marketers know that it is not wise to try and be all things to all people. Instead, they aim at market segments. When the laboratories of Lever Brothers discovered how to make detergent into solid bars, they had invented what I, in the context of this note, call a device. David Ogilvy (of the advertising agency) considered the choice of aiming a product at men with dirty hands or at ladies with dry skin. Each of these alternatives defines a product. A different product. In its way, 'Men with Dirty Hands' is an invention. So is 'Ladies with Dry Skin.' As it happens, only one of the products, the latter, was realized. Only when the 'product' was invented and was selected for 'production,' did it receive a name: DOVE. The rest is history [2].
Even a scientist exploits this kind of invention. Were you ever in a position where you had plenty of 'material' (theorems, observations), but had great trouble writing it up? You had the equivalent of a device, but did not know what to do with it: you still needed to invent a product. The difference between a notebook with theorems or observations and a paper that makes an impact is an invention similar to the one that gave the world DOVE. A successful scientist, even one who never comes near to any application, let alone commercial exploitation of his work, is good at a form of marketing. His papers have to be products if they are to make an impact. If it is too hard to base a product on the material, then it may be that the theorems or observations are not worth knowing.
I view LP as an invention similar to solid-state electronics: a device with enormous potential. I view Prolog as analogous to the early Transistor. It is an experimenter's kit, useful in a limited way, but only just starting to exploit the potential of LP. The analogs of the portable radio and the hearing aid are still to come. They will only come about as the result of additional inventions.
In what direction should we look? One possibility is the adaptation of existing inventions, such as spreadsheets [4] and relational tables [5]. Another one is to take Prolog as starting point and ask for what applications it is superior to other languages, without any doubt. I am afraid the answer is 'none', although Prolog is superior, in a general, rather vague way. There are certain application areas where Prolog almost makes it, for example:
[2] David Ogilvy: 'On Advertising,' Wiley, Toronto, 1983.
[3] M.H. van Emden: 'Rational reconstruction in computing,' unpublished memorandum vE-89-28.
[4] M.H. van Emden, M. Ohki, and A. Takeuchi: 'Spreadsheets as a subset of LP', New Generation Computing, vol. 4 (1986), pp.287-304.
[5] M.H.M. Cheng, M.H. van Emden, and J.H.M. Lee: 'Tables as user interface for logic programs', Proc. Int. Conf. Fifth Generation Computer Systems 1988, pp.784-791.
[6] H. Gallaire: 'Boosting LP', Proc. 4th Int. Conf. on LP, Jean-Louis Lassez (ed.), MIT Press 1987, pp.962-988.
M.H. van Emden Dept. of Computer Science Univ. of Victoria Victoria, B.C., Canada Email: vanemden@csr.uvic.ca
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |