The annual lectures of which this is the first are intended to commemorate David Glass.footnote He was my friend, and the friend of others in this room who don’t need this occasion to recall him in the presence of his inseparable partner, Ruth Glass. He was also one of the most distinguished scholars to teach at the LSE, with which he was so long associated and whose reputation owes much to his presence there. I might add that he represented its finest traditions at a time when not everyone there did so: the traditions of understanding society in order to make it better, of an instinctive radicalism, of an institution whose students, like himself, were not born with silver spoons in their mouths. It is typical that he concluded his very first book on demography—of which he was in his lifetime the most eminent practitioner in Britain—with the call to ‘provide conditions in which the working class is able to bring up children without thereby suffering from economic and social hardship’. He was proud to be the first social scientist to be elected to the Royal Society since the great Dr William Farr in 1855, because he saw himself (like Farr) as a social scientist in and for society, and not just about society.

So it is natural that the lectures devoted to his memory should be about ‘social trends’, which I understand to mean in the broad sense the enquiry into the direction of social development and what we can do about it. That implies looking into the future, so far as this is possible. This is a risky, frequently a disappointing, but also a necessary activity. And all prediction about the real world rests to a great extent on some sort of inferences about the future from what has happened in the past, that is to say from history. The historian ought therefore to have something relevant to say about the subject. Conversely, history cannot get away from the future, if only because there is no line which divides the two. What I have just said now belongs to the past. What I am about to say belongs to the future. Somewhere between the two there is a notional but constantly moving point which, if you like, you can call ‘the present’. There may be technical reasons for considering past and future differently, as any bookmaker knows. There may also be technical reasons for distinguishing present from past. We cannot ask the past for direct answers to any questions which have not already been put to it, though we can use our ingenuity as historians to read indirect answers into what it has left behind. Conversely, as every pollster knows, we can ask the present any answerable question, though by the time it is answered and recorded it will also, strictly speaking, belong to the past, albeit the recent past. Nevertheless past, present and future form a continuum.

Moreover, even when historians and philosophers want to make a sharp distinction between past and future, as some do, nobody else will follow them. All human beings and societies are rooted in the past—that of their families, communities, nations or other reference groups, or even of personal memory—and all define their position in relation to it, positively or negatively. Today as much as ever: one is almost tempted to say ‘more than ever’. What is more, the overwhelmingly large part of conscious human action which is based on learning, memory and experience, constitutes a vast mechanism for constantly confronting past, present and future. People cannot help trying to forecast the future by some form of reading the past. They have to. The ordinary processes of conscious human life, not to mention public policy, require it. And of course they do so on the justified assumption that, by and large, the future is systematically connected with the past, which in turn is not an arbitrary concatenation of circumstances and events. The structures of human societies, their processes and mechanisms of reproduction, change and transformation are such as to restrict the number of things that can happen, determine some of the things that will happen, and make it possible to assign greater or lesser probabilities to much of the rest. This implies a certain (admittedly limited) range of predictability—but, as we all know, this is by no means the same as successful forecasting. Still, it is worth bearing in mind that unpredictability looms so large mainly because arguments about prediction tend to concentrate, for obvious reasons, on those parts of the future where uncertainty appears to be greatest, and not on those where it is least. Meteorologists are not needed to tell us that spring will follow winter.

My own view is that it is desirable, possible and even necessary to forecast the future to some extent. This implies neither that the future is determined or, even if it were, knowable. It does not imply that there are no alternative choices or outcomes, and even less that forecasters are right. The questions I have in mind are rather: How much prediction? Of what kind? How can it be improved? And where do historians fit into this? Even if anyone can answer these questions, there will still be much of the future about which we can know nothing, for theoretical or practical reasons, but at least we may concentrate our efforts more effectively.

However, before I consider these questions, let me reflect for a moment on the reasons why the function of prognosis is not only so unpopular among many historians, but also why so little intellectual effort has gone into improving it, or considering its problems, even among those historians firmly committed to its desirability and practicability, such as Marxists. The answer, you may say, is obvious. The track-record of historical prediction is, to put it moderately, patchy. Everyone of us who has made predictions has frequently fallen flat on his or her face. The safest thing is to avoid prophecy by claiming that our professional activities stop at yesterday, or to confine ourselves to the studied ambiguities which used to be the speciality of ancient oracles and are still the stock-in-trade of newspaper astrologers. But in fact, a poor predictive record has not stopped other people, disciplines or pseudo-disciplines from forecasting. There is a large industry devoted to it today, undeterred by its failures and uncertainties. The Rand Corporation has even in despair re-established an up-dated version of the Oracle of Delphi (I am not joking, the name of this peculiar game is the ‘Delphi technique’) by asking selected groups of experts to consult their chicken’s entrails and then drawing conclusions from such consensus as may or may not emerge. Moreover, there are plenty of examples of good predictions among historians, social scientists and academically unclassifiable observers. If you do not wish to have Marx quoted at you, let me refer you to de Tocqueville and Burckhardt. Unless we assume, what is unlikely, that these are purely random hits, we must accept that they are based on methods which are worth enquiring into if we are to concentrate our fire on targets we can expect to hit and improve our ratio of bulls’ eyes to misses. And, conversely, the reasons for notorious flops are worth enquiring into with the same object.