Not Facing the Future: Computing Chaos in the New Century?

When the first computers were built, memory and processing time were both tremendously expensive. Trying to imagine how much more expensive than now is exceedingly difficult. A useful technique to save a lot of money was to reduce the length of all dates so that instead of 14.07.1997 (ten characters) only the last two digits of the year were used, and no separators: to give 140797 (six characters - a 40% saving, for no effective loss of information at that time).

It was, accurately, assumed that the price of memory and processing would fall over time in such a way that the cost of processing the full date would be insignificant well before the end of the century. Therefore, it was, reasonably, assumed that the programs that were written with abbreviated dates would not still be in use when the change of century came around, which would be when the 'purely theoretical' loss of information would become a practical loss of information. Now we know that they will be. This is because in the mean time, huge numbers of other programs have been written using the program code that was designed in these early days as a starting-point in one form or another. There has never been a thorough sorting out of which lines of program code use abbreviated dates. One common route for perpetuating the problem has been that programs to help in the writing of other programs themselves were written on the assumption that only two digits were needed for the year. This has continued until remarkably recently (Seidel, 1997), and may still be continuing. It is quite possible that new computer programs are being written that will not continue to work properly when the century changes: in the jargon, they will not be 'millennium-compliant'.

Dates are used in huge numbers of ways in computing systems, many of which are not obvious, and hidden under no end of cryptic and whimsical names. It will not be quick or simple to check all of the ways in which they are used in programs. It cannot be done completely by automated tools, and it is likely that there will be some programs that it will never be possible to check at all automatically. Even when automated checking is possible, there is no guarantee that an automated check has found all of the uses of dates that are there, because to a computer, a date often looks just like any other number.

For some computer systems, the onset of the turn of the century will cause the programs to behave in unusual ways. A bank's computer may well be programmed to close accounts that have not had any transactions for ten years (North, 1997). When the date becomes 01.01.00, the computer might check an account, and discover that there have been no transactions in the (non-existent) years -10, -09, -08, -07, -06, -05, -04, -03, -02 and -01 and decide to close the account. It might do this for all the accounts held with that bank. While the staff at the bank might realise that the program is not behaving how they would wish it to, and thus the staff will work to sort out the consequences, customers might be unable to use their accounts for a few weeks - they might be without money for a few weeks. If, that is, the bank's computer is not millennium-compliant in this particular way.

As the millennium changes, the ordinary computer user may find that their PC behaves oddly, if they have a PC (I believe Apple computers may be better able to cope with the change of date - see Apple Computer, Inc., 1996) but there are at least five good reasons why individuals should not rush to replace their PCs now (NOTE: THE FOLLOWING DOES NOT APPLY TO BUSINESSES).

Firstly, there is a spectacularly high rate of obsolescence and replacement in personal computing already. A fair proportion of hardware and software would get replaced over a time-span of two and a half years, anyway.

Secondly, there is a high degree of uncertainty about precisely which systems and combinations of systems will fail. This has two meanings: you do not know how much of your system will fail (or whether it will fail at all), and you cannot be sure that a new system which claims to be millennium-compliant, will actually be (Greenwich Mean Time, 1997), until you comprehensively simulate the change of century, or experience it for real, with all of the hardware and software you will actually have when the millennium turns actually installed. In one test (Greenwich Mean Time, 1997) 47% of suppliers of software that could have trouble when the year 2000 starts "claimed that these products were year 2000 'compliant', 'safe' or 'ready'."

Thirdly, most individuals can survive even if their PC is out of order (albeit with a little inconvenience).

Fourthly, hardware manufacturers already churn out great quantities of pollution (Corporate Watch, 1997) to meet a demand that they and the software houses work so hard to generate already, and they have a vested interest in exaggerating the effects of the change of century on personal computing to generate a marketing opportunity. As Asia Pages puts it: "although in reality, many stand-alone machines may not feel the impact of the [millennium] bug because the applications they run are not dependant on the 'year 2000' issue. ... companies [and] home owners ... are likely to want such fixes in order to feel 'safe' - lots of money here".

Finally, there will almost certainly be software available after the turn of the millennium to get many PC systems working properly again, if many need it.

These five factors suggest that the private individual is probably better advised to not buy a new computer system until after the turn of the millennium. There is little advantage to individual PC owners in replacing perfectly good hardware and software, just to get some that is claimed to be millennium-compliant, but which may not actually work properly when the year 2000 turns. Businesses, however, may be bankrupted if they follow this course: for businesses, the third reason (above) most definitely does not hold.

At the heart of the problem of the turning of the century could well be the huge number of safety-critical systems used by the military and industry. For very many of these we do not know whether there is a problem caused by the turn of the year 2000.

Amongst the worse problems could be that "Date sensitive weapons systems could fail" (CLB Enterprises, Inc, 1997). Does this mean that they will not work if the military feel they need to use them, or that they go off automatically, though? Most people would not be happy with either prospect.

Computers are used to control all sorts of industrial processes, some dealing with lethal chemicals. Some of these may cause immediate problems if they are not millennium-compliant, while others which control maintenance will only cause trouble after some time: by not reminding operators that they need maintenance, they could fail catastrophically several years into the new millennium.

The major problem here isn't so much that we know that systems will fail, as that we cannot be sure that they will not until they are properly checked for millennium-compliance. Checking is the problem. Many computers are 'built in' to industrial plants in such a way as to mean that it is not obvious that they are even there, as embedded chips, and parts of the control system (Travis, 1997). They do not have keyboards and visual display units. Even if it is realised that a computer chip is there, it is almost certain that nobody at that site has records of what software has been built in - let alone whether that software is millennium-compliant. It is quite possible that nobody even knows who supplied the software. Trying to take a look at the lines of the software code that are built in could be technically very difficult and even if it was done, it would be expensive.

Yet failure to check could result in releases of poisons into the air and into water with dire consequences for huge numbers of people and other species. All told, "There is a high probability that deaths and major disruptions could result from the malfunctioning of the noncompliant computers." (CLB Enterprises, Inc 1997)

Note: added January 2000 - Thus far, at least, it seems either these fears were exaggerated, or that failures due to warning systems failing have not yet occured. If it is the latter, there is still time for action to prevent problems

Meanwhile the cost of checking is increasing as an increasing number of businesses who have realised that they need to act to prepare their business for the turn of the century are chasing a pool of computing and software professionals that is not growing very quickly. Some predictions are that the price of checking will quadruple before the end of the century (Patrizio, 1997).

There is a temptation, against this background, to resist spending large sums of money 'just to check' safety-critical systems when everything might be all right anyway. Such an attitude is understandable, but irresponsible.

Ensuring that safety-critical systems are millennium-compliant should be among the highest priorities for society and thus for governments. There is room to argue that responsible governments should pass laws to make sure that ensuring that safety-critical systems work should be given priority, given the shortage of programmer-hours before the 'millennium bug' needs to be fixed.

Some of those who warn us about the problem of the year 2000 claim that it could result in there being huge disruption to banking. In actual fact (Estes, 1997) large proportions of the banking industry in both the UK and the USA, and presumably in other countries too, is well on target to be millennium-compliant in time.

The doom-sayers (eg North, 1997; Mills, 1997) are worried that the proportion of the banking world that will not meet the deadline for this project will cause so much disruption that society as we know it will cease to function. It is hard to know for sure, but many banks are spending a lot of money to get ready for the change of century. My own guess is that there will be some inconvenience, and some disruption, but life will return to a sort of normality, if the problems in the financial sector are the biggest we face.