The (Nested) Education Vector

Transcribed from the printed issue; not yet reviewed. Read it in the PDF of the issue, page 18

Hurrah for a VECTOR dedicated to Education - how nice it would be to see Education rushing en masse to subscribe to VECTOR! Perhaps the best route to this goal is through statistics. In a note elsewhere in this edition, Peter Kelly from Newcastle University and I point out that there is no other high level language which provides a regression package free with every interpreter supplied!

In the closing address of the APL 84 conference in Helsinki, Fred Perkins of I. P. Sharp pointed out that computer users were increasingly package users, and as such have no reason for partiality or loyalty towards one underlying high level language rather than another. The success of APL therefore depended on the quality of the packages which were written and marketed by APL people; to carry on parading the merits of the APL language whilst ingoring this fact of marketing life was to hasten the doom of the very thing they were trying to promote.

There is however a cycle in the ways of computer usage in which the ever more sophisticated package user eventually becomes shackled by the restraints imposed by the software whose convenience gave him such delight not so long before. At this point there is no answer but to retreat again to the base language. In no subject area is this more true than in statistics, and it can be fairly said that APL must be the most flexible statistics package around.

Right at the start of statistical computing it is there with facilities for calculating simple descriptive statistics with ease equal to that of any of the popular packages. As noted above this even extends to multivariate regression, and if convenience of data-entry is taken into account, APL must surely come out as the top package at this level.

As statistical specialism increases it is undeniable that the speciality languages such as GLIM provide powerful algorithms on tap which are a rich and convenient source of ready-made number crunching, whose appeal outweighs the time required to provide the equivalent programs in APL. On the other hand the range of worthwhile statistical techniques is increasing at an exponential rate, and the special packages such as those mentioned can only contain a finite number of them, and that pre-determined at the time of last release. This is where the frustrated statistical user must surely be weaned back to APL. APL is there at the beginning and there at the end, the alpha and omega for the serious statistician with a computer.

So where does the education begin? Surely at the earliest introduction to statistics, be that at school or university. The case for the mathematics and statistics department having I-APL as a matter of routine is enormously strong, and we hope to see micro Labs up and down the country in which I-APL is as commonly available as spreadsheets and wordprocessors. By the time you read this I-APL will be available for PCs, and for BBCs not long afterwards. The benefit which can accrue from having individual members implant the idea of APL in the minds of their teacher friends is potentially enormous - personal recommendation is much the best advertisement. And if the response is "If APL is so good, why has't my Professor/LEA Adviser/Computer Science Department told me about it?" be quick to say "It's only now that you can afford it, so why settle for less than the best!"

Finally, and announcement which will be of great interest to APL enthusiasts in Universities and Polytechnics. An APL2 service is now available to them . . . for FREE! British academics who wish to try APL2 at no expense to themselves can now do so by applying to Warwick University. Withs IBM's support, Warwick have now got the ability to make their internal APL2 service available over JANET. Please announce this facility to any academic friends who may have been wanting to try APL2.

Detailed information can be obtained from:

Mike Hunt
Director of Computing Services
University of Warwick
Coventry CV4 7AL
Tel: 0203 523355