Envisaging the latent possibilities of electronic computers, even sober scientists are apt to indulge in daydreaming. If we are to believe some of them, information machines will be able in the near future to do everything that human beings believe they are capable of doing: waging war, drawing up new tax laws, and making films for television - maybe even inventing a shoestring that won't break.
We are unwilling to lose ourselves in Utopian visions. Let us see what electronic computers are actually capable of doing today, or what - if the programmers and technicians are given another few years to play with them - they are certain to accomplish.
Let us begin at the point at which we have already seen computers at work: in the office. Electronic automata for business use have been in production since 1953. In their construction, they are among the simpler types. They are not expected to perform miracles of calculation. They are asked to add and multiply and to calculate interest. Above all, they need big storages so that they can retain particulars of accounts and stocks in hand, itemized prices and discount rates, pay scales, working hours of employees, periods of delivery, and tax-exempt percentages for depreciation, in their heads.
You will be able to see even from this brief list that a commercial computer can replace entire departments of a business. In fact, it has to be in a position to do so, for it costs several hundreds of thousands, and the programs with tens of thousands of instructions, which must be worked out afresh for practically every firm, swallow up a few tens of thousands more.
Even more expense may be incurred if the staff does not treat the computers as they have a right to be treated. Electronic computers expect you to give them all incoming invoices and outgoing packing slips, details of taxes due for payment, bills, and so on, on punched cards or in some other form - but all of them. Experienced bookkeepers and clerks know what to do if an invoice or stock control record card which should be in its proper place is suddenly not to be found. But a computer is not attuned to such evidence of human fallibility. In almost every electronically equipped business, the trouble starts when some very junior clerk is preoccupied with thoughts of the world skiing championships and forgets to record the day's withdrawals from and additions to the warehouse - and when no human purchasing manager is there with years of experience to let him know that something must be missing. Once the automatic calculator has started counting with wrong numbers, it can cost a great deal of money to put it on the right track again.
Another problem is the childlike faith owners of computers are apt to have in the infallibility of their newly acquired robots. If a computer has been delivering absolutely exact information for three months, they imagine that it will go on doing so, unchecked, for all eternity. They forget that even a computer is the work of human hands, in other words, that it is apt to display sheer cussedness. Now it forgets something (an event which in itself may not have disastrous consequences, for checks are built in, or programmed in) or it may break down. If that happens, a company may go to wrack and ruin. Big businesses, therefore, often buy two computers, one of them to keep in reserve.
A Swiss paper some years ago warned medium-sized businesses not to allow themselves to be bewitched by the will-o'-the-wisp of progress, but to make a precise calculation beforehand to find out whether the purchase of a computer would really be a paying proposition as far as they were concerned. In many cases, it may certainly look at first glance as if the computer - if its price is compared with that of a dozen accountants - is not earning its keep. But it may well have advantages nonetheless, because it keeps the company's books constantly up to date - up to the minute, in fact - while accounts and specifications kept by hand must always necessarily be at least three to thirty days behind. In the case of one big American company it is known that its computer ate up more money than it saved in salaries for human employees. But the company worked closely with the United States Government; with the computer, the firm could settle its accounts two weeks earlier and receive the payments due to it that much earlier too. So the bank charged less interest and the machine saved money after all.
Hard-working computers not only keep a company's books in order. They remind their owners of the due dates of bills. They will even type these on teleprinters and take them, ready to mail, to the machines that will put them into envelopes and stamp them. They not only observe that the quantity of X-117 steel bolts in inventory has fallen dangerously low; they write the order at once for a new supply. They not only note that thirty thousand new catalogues have just been delivered and that the printers must be paid; they give instructions at the same time to the addressing machine to print an envelope for every customer so that a copy of the catalogue can be sent him. (Sometimes, unhappily, they may forget to give enough customers' names to the addresser; some years ago an American sheep farmer had a thousand or so catalogues from a mail order firm delivered to his house - and his front garden.)
Aside from such activities, the same computers are capable of performances that seem to be marvels of scientific reasoning. At night, when all self-respecting people - and the accounting department - are asleep, the information stored in its memory enables the computer to compile yards of statistics, work out the profitability of the various departments, draw up market analyses and rejoice the heart of the company's president when he arrives on the morrow by presenting him with a thoroughly worked out diagnosis proving that he is running his business crazily.
Just as they are conquering commercial and industrial enterprises, computers are invading banks and insurance offices, where they arouse astonishment, praise and bitterness. They keep all accounts automatically, calculate interest before clearing and after, determine premiums and surcharges, and make it very hard indeed even for the most experienced bank official to embezzle money. But the fact that there are ways and means of cheating a computer all the same was demonstrated by a bank director who made a close personal study of the art of programming for this special purpose and kept the computer busy with all kinds of minor transfers and retransfers of cash from one account to another, taking good care to see that, while all this was going on, a tiny percentage accrued on each occasion to his own account. As the bank's auditors were themselves not very well versed in the laws of electronic bookkeeping, it was a long time before they caught up with the director, who was then with his second program behind him.
Banks and commercial businesses, by the way, are very interested in a process which makes it possible for computers to read handwritten numbers and letters. In this procedure, which will probably be perfected soon, the computer compares handwriting, which is broken up into electronic impulses as in a television transmission, with written examples it has previously been taught.
Electronic computers are naturally ideal for assembling vast masses of figures so cunningly that bookkeeping, compared with the result, is as thrilling as speculating on the Stock Exchange: for statistical purposes. Great electronic computers evaluate the multitudinous items of information contained in filled-out census blanks, calculate how many independent businessmen in Birmingham have cars with engine sizes of less than 1000 c.c. and how many fifty-year-old Post Office Officials live in Plymouth. Most people are not very interested in such odd bits of information, but the Post Office, which wants to know how much money it will need for pensions payable fifteen years from now, thinks differently. In the old days it was impossible to extract such marginal information from a census, however much the Post Office or anyone else might have been interested in it, for no one had the time to ferret it out.
Government departments, even those which cannot normally be suspected of an inordinate love for ultra-modem methods, tend to be more and more fascinated by the possibilities electronic computers offer them. An Economy Committee in one city suggested to the administration that it might be a good idea to buy a computer to work out the pay of officials and other employees. The Committee pointed out that the machine would calculate the salaries and wages of all the government's employees, for a mere quarter of a million a year. A quarter of a million is a lot of money for merely counting out the contents of wage packets. But the old method had cost the city government two and a half times more than that every year.
The machines mentioned so far in this chapter are "general purpose" installations. They operate with practically any program given them, and they are universally utilizable. The manager of the business owning them could, if he wished, use them to evaluate a census, to precalculate the dimensions of a bridge span, or to play chess. "Universally programmable" installations cost around $500,000 or $750,000 (£180,000 to £270,000). Additional units, reserve storages and so on can raise the price to $2,000,000 or $2,500,000 (£715,000 or £.900,000). For the money, however, you have a beautiful contrivance, versatile, adaptable and with an appetite for any kind of program, produced by happy engineers - as advertising people might say.