In what follows we neglect the entertainment functions of the media and concentrate on their news functions. We neglect sport, which sits on the fence, but we do allow the rather irresponsible dressing up of entertainment as news.
A story
Suppose we are reading a story A in website B (perhaps Facebook), filed by C (perhaps a journalist claiming allegiance to a well known newspaper).
A goes something like:
Opening remarks by C.
According to D (perhaps some think tanker),
E (perhaps Teresa Corbyn) says that,
F1
F2
F3
etc
F23
Concluding G (perhaps that we should renationalise the coal mines before they are all flooded out).
Closing remarks by C.
The point of the story might be to promote the conclusion G, to rubbish the conclusion G or something else altogether. G might be incidental. Perhaps the real point is to rubbish Teresa. Perhaps to make C sound well informed and important. Perhaps something which is nothing to do with the ostensible subject of the story, perhaps to advertise product H, mentioned seemingly incidentally in the story, perhaps advertised next to the story.
We understand that stories in newspapers or on websites are not usually organised in quite this way, but we do believe that they often amount to something along these lines and the present pretence serves to make the point developed below that the apparently logical organisation of these stories is misleading, is indeed a pretence.
The actors and their objectives
Thinking about the point of the story leads on to the more general point that the various actors in this story – B, C, D and E – all have different objectives and motivations.
B may only care about the number of advertising revenue generating clicks and have no interest in the actual content. If rubbish and worse generates clicks, fine so long as it is not illegal. Never mind about the spirit of the law or the public good.
One might think that the filer C has lead responsibility for the content. Does one know and trust him? Does one know anything about him at all? What about the use of pseudonyms? Do respectable papers like the Guardian allow their journalists to fly under false colours?
In any event, the way that A is put together means that C has plenty of scope to dump any blame going on D or E. No responsibility here, thank you very much.
We might know something about D and E, we may know and trust them, but they may know nothing about story A, not recognise it at all. Words might have been put into their mouths. On the other hand, either one of them might have given C the benefit of a private briefing, for undeclared reasons of their own.
In the olden days things were much simpler, and many of us gave our allegiance to a newspaper. We trusted what we read in it, particularly if the piece in question was by a journalist we knew, at least in print. Both newspaper and journalist were respectable and it was unlikely that they would collude in the production of a lie. They might have their slant, but they would draw back from outright lying or misrepresentation. Indeed, in those days, newspapers used to reproduce chunks of the debates in our Houses of Parliament, on a daily basis, rather than dishing out heavily edited dollops, third or fourth hand. But we should not overdo the good old days: there have always been bad newspapers and irresponsible newspaper owners.
The rules of a story
Each of the twenty three F’s above is either the statement of a fact or the deduction of a fact, using a rule of inference, from the facts which have gone before. An example of applying a rule of inference to a couple of facts might be ‘Amazon supplies half the books sold in Scotland. There has been a serious fire in their one and only warehouse up there. Therefore Scotland is going to be short of books for a few weeks’. We are supposed to be happy to go along with the facts which are stated. Perhaps we knew about them already, perhaps their truth is self-evident or perhaps we are prepared to take them on trust from this reliable source. And we are supposed to be happy with the application of rules of inference to those facts, happy that those rules have not been abused. The whole thing hangs nicely together and drives us to the conclusion G. Not sheep to the slaughter at all; rather sensible and knowledgeable voters being put in the picture.
A sequence which has the structure, the appearance of a proof in mathematics, with the important proviso that the whole thing is much looser and takes a lot more on trust.
So an example of a statement in such a proof might be something nice and simple like: ‘Teresa Corbyn is a liar’. Which might mean that Teresa lies all the time, although the strongest version ‘everything she says is a lie’ is rather unlikely. It might mean that she has been caught saying things which turned out after the event to be untrue, not necessarily the same thing as lying in the ordinary sense of the word. It might mean that the reporter believes that Teresa is lying on this particular occasion, about this particular fact. It might mean no more than the reporter just wants to dent our trust in Teresa, without offering any particular reason. It might also be the case that the attribution is unfair, that Teresa never uttered this particular fact, at least not in the way suggested by the report. All in all a long way from a statement like ‘the triangle ABC is congruent to the triangle DEF’.
Another example of a story might go along the following lines. There has been a murder of a young child in some town in the heart of England. Various statements about what a lovely child and what a lovely place. Various statements about the gruesome murder. Various statements about the grieving family. Various statements about the bad past life of the perpetrator. Question: why should we spend our hard-earned money on the perpetrator? Conclusion: jail too good for him. String him up!
The filer of such a story may well not care one way or the other. The murder is just a vehicle for a bit of purple prose to keep his readers entertained for a few moments. The story works by cranking up our anger. The question then makes an implicit appeal to the rule that says that we should not reward bad behaviour. The conclusion is then the only alternative. Which also has the merit of appeasing the anger, settling the blood-debt vicariously. A story which pays scant attention to the rules set out above; a story which is not much of an argument, never mind a proof.
The western, liberal ideal is that we read stories which are self sufficient. We know about and believe the facts, we can follow the rules of inference. The story stands up and we agree with the conclusion. It does not matter how many layers of journalist and commentator the story and has come through and whether or not we trust them. The story stands up of itself.
Part of this is the national statistical institutes run by most countries. Institutes which produce facts which can be relied on; which can be included in arguments without much comment.
By way of comparison, we also have the framework provided by laws. We don’t have to worry about whether the law is right or not, someone we trust has already done that: we can just stick to the laws while going about our daily business.
The difficulty is that most of the facts in arguments about public affairs are as tangled as the liar fact above. And the inferential links between them are tangled too, often more or less non-existent as in the murder story above. Which means that what we have, despite appearances, is nothing much like a proof in mathematics and while we might feel that an argument was plausible, we would not go so far as to say that it was true or false. We are much more likely, or at least we should be much more likely, to say that the argument is plausible, even very plausible. We then want to back that up by our assessment of the source. How much trust can we put in that?
And we are prepared to believe a plausible story from a trustworthy source.
The present case
In the present case we have the possibility that computers are writing the stuff that we read in the papers or on the Internet. With the article in the Guardian suggesting that this might include all kinds of plausible, but invented, quotes from all kinds of more or less respectable and trustworthy people.
Now while the writers of this particular software may have decided not to release it for the time being, they are not going to stop it. The genie is very much out of the bottle and we have to learn to live with it - it being a substantial increase in the number of fake stories floating around on the Internet.
Without giving the matter the time and energy it deserves, my thought is that part at least of the answer lies in promoting three ideas. First, we must give up on the idea that we are in any position to check up on most of the stories that we read: we have to take them largely on trust, trust which needs to be built and maintained. Second, that if we want real news rather than garbage, we have got to be prepared to pay for it. The primary product of the newsagent must be news, not advertising revenue generating clicks. Third and last, that we should be buying our news from a respectable source. Respectable includes having visible governance and finance, having track record and having visible journalists. The journalists might use pseudonyms, but they must be real people whom we can get hold of and interrogate, should need arise. We must know enough about them for it to be reasonable to take them on trust. And speaking for myself, I never cared for the practise of giving criminals a platform in newspapers – thinking here of one particular criminal who was once writing regular pieces for the New Statesman. A criminal, even an apparently reformed criminal, might be a source of useful information, but is not a person in whom I would care to put my trust.
It is true that Wikipedia, partly because their (unpaid) editors do not want unpleasant emails about their work for Wikipedia cluttering up their personal email accounts, allows those editors to be anonymous, certainly as far as the public is concerned. We have to put our faith in the Wikipedia system being able to weed out the bad apples. But for matters of public policy, I do not think that that is good enough. If you want to pronounce, for example on the legalisation of recreational drugs, you have to have a name and you have to have an existence which can be checked out. Stories, despite the western liberal aspiration mentioned above, are not self sufficient and we need to be able to put trust in the story teller.
I note also that Wikipedia articles usually come with references, many of which can be taken up over the Internet. This is fine for relatively serious study, but not to fine for items in newspapers or floating around on social media, where you are not usually going to take the trouble to take up references, even supposing there were any. Stories are supposed to be self contained.
Conclusions
I do not think that this particular genie can be kept in the bottle, at least not for very long, short of going down the authoritarian road of, for example, the Chinese, with their state control of the media.
Instead, we have to try to move towards a world of information and chatter which remains available to all, but which also contains responsible and respectable players in whom we can put our trust, when need arises. Which, to coin a phrase, might manage to square the circle.
Part of this has to be educating the public at large, getting the public to understand that there is no free lunch, here or elsewhere. And part of this has to be government prodding, to get these responsible and respectable players into place; players which are likely to be of the same sort of size as our present Guardian and Economist, rather than the size of the likes of Google and Facebook. Or, government failing, do we have to look to some billionaire to put some of his billions into some appropriate trust?
I close by noting first, that respectable players might well come to include respectable computers. There might well come to be computer programs which we trust to assess the facts better than we could assess themselves ourselves. Rather as we already have medical computer programs which can, in their field of expertise, do better than most real doctors. And second, that though we might knock reason and reasoning, it is what we have, it is an important part of what has made us what we are. It is a more reliable compass through life than emotion, bluster and intuition, so let’s make the most of it.
PS: today’s (print edition) Guardian has a front page headline about the need to do something about the digital gangsters operating Facebook. The people at the Guardian have clearly worked out that we need a break from our diet of Brexit. But see reference 4.
References
Reference 1: Big brother is reading you: why AI text generator may be too dangerous to release - Alex Hern, The Guardian - 2019.
Reference 2: https://openai.com/. But the Wikipedia article about them might be a better place to start.
Reference 3: https://psmv4.blogspot.com/2019/02/mercier-sperber-review.html. Where we talk about a book which talks about reasoning. In this context, a useful reminder that the reasons we give for things are often very feeble and that our thinking is often warped by undeclared biases and prejudices.
Reference 4: Disinformation and ‘fake news’: Final Report - House of Commons: Digital, Culture, Media and Sport Committee – 2019. HC 1791. 109 pages of pdf text, 51 conclusions and recommendations, handily listed at the end. How much of it will I ever get to read? What little I have read so far, suggests that the committee wants to address the supply side of the question, for example by beating up Facebook. Which is useful, which helps, but which, to my mind, needs to be supplemented by more work on the demand side, on educating the public.
No comments:
Post a Comment