Francis Bacon, knowledge, power and fakes

One of my favourite phases in my long-abandoned academic career was the few years I spent studying the history of physics and of the scientific method - where I was able to dive deeply into the unsorted, various and frequently barking-mad works of an astonishing man, who lived and worked within a mile of where I grew up: Sir Frances Bacon.

For those unfamiliar with his work: Bacon, 1st Viscount St Alban (1561 – 1626¹), was a major player in the court of James I, and was interested and involved in absolutely everything to do with the new sciences that were emerging at the time, as well as how it could be put to use to expand the power of the British state. He is probably best-known now as the figure who is said to have first formulated what we now call the scientific method. And he is gloried for this achievement - memorials to him are not confined to St Albans where he was very much Lord of the manor - this picture is his statue in the Library of Congress.

In fact, when you read what Bacon actually wrote (or more often, what he claimed he would write, forthcoming in the next volume, etc, etc), it’s far, far weirder and more interesting than some dry and precise formulation of steps of something that we can recognise as an early scientific method.

His Novum Organum and the incomplete Instauratio Magna are indeed of dizzying ambition - an organisation of all the emerging sciences in order to fulfil man’s destiny master of nature. But what they actually end up as is a woolly and impractical assortment of rules of reasoning: some dizzyingly insightful, others clearly inspired by alchemy and the occult, and still others which just seemed to occur to him as good ideas at the time. It’s a glorious, wonderful mess of thinking from a man with one foot in the pre-scientific age, and the other stepping forward on a new path. And because the man was slightly nuts - this path occasionally takes a diversion to talk about things like different blends of pitch for five pages before petering out. It’s not the scientific method, or anything close to it, but it is wonderful nonetheless.

Computers and the internet has greatly aided research on Bacon and his near-contemporaries. Bacon’s fragmentary and incomplete works are hard to find in print, even more difficult to make your way through in a thematic way, or to understand references to near-contemporaries like Gilbert. But the complete works of Bacon are now available (and searchable) here, so scholars have an invaluable source at their fingertips.

You would have thought that Bacon himself would have approved of such technology for spreading his works. The originator of the aphorism² “Knowledge is Power” was - after all - conscious of the importance access to authoritative sources of knowledge, and held out the printing press as one of the three inventions of his own time that were changing the world, alongside gunpowder and the compass.

An instance of the ironic

However, earlier this year, Professor Tyler Cowen - holder of the Holbert L. Harris chair in economics at George Mason University - brought out a strong irony in an article on the Marginal Revolution blog entitled Who was the most important critic of the printing press in the 17th century?

Professor Cowen pointed out that Francis Bacon followed up his praise for the printing press with a series of warnings - for he felt that the press had become a force for evil that even outweighed its use for progress. And these could be thought to apply even more strongly to the internet. Cowen’s article makes four main points:

– First, he argued that the printing press had flooded the world with too many books, especially of ancient and scholastic learning, that were either obsolete, irrelevant, or misleading for the pursuit of true knowledge. He compared the proliferation of books to the confusion of languages at the Tower of Babel, and lamented that “the multiplication of books is a burden of the world” (Bacon, 1605, Book I, Chapter VIII, section 2).

– Second, he argued that the printing press had encouraged the idolatry of books and authors, as people tended to accept the authority of printed texts without examining their merits, sources, or methods. He accused the printing press of “making the world lazy with books” and of “putting a kind of reverence and religion into letters” (Bacon, 1605, Book I, Chapter VIII, sections 3 and 4).

– Third, he argued that the printing press had hindered the advancement of learning by promoting the imitation and repetition of old opinions, rather than the invention and experimentation of new ones. He claimed that “printing has made the world more set and less inventive” and that “men have come to be as it were a kind of fungous generation, not generating from within by the force and virtue of the mind, but growing upon one another and propagating from without by the way of tradition and authority” (Bacon, 1605, Book I, Chapter VIII, sections 5 and 6).

– Fourth, he argued that the printing press had fostered the dissemination of false, frivolous, or harmful knowledge, such as superstitions, fables, libels, and scandals, that corrupted the minds and morals of the readers. He cautioned that “printing has been guilty of much wrong; for by it many false and vicious things have been propagated in the world” and that “it is a thing more subject to the humours and passions of men than perhaps any other” (Bacon, 1605, Book I, Chapter VIII, sections 7 and 8).

(Cowen, February 2023)

Overall, Professor Cowen concludes that Bacon is not condemning the invention of the printing press itself, but instead the way it was being used. With good practices and regulation, Bacon believes balance could be restored in favour of its good use. Cowen - a libertarian economist - notes that Bacon’s critique was and is controversial, but presents it as a product of its time as new attitudes to knowledge arose.

However, there’s a flaw in Professor Cowen’s analysis: every word is a lie.

An instance of the bullshitter

None of the material Cowen refers to appears in any of Bacon’s work. None of the direct quotes, nor any of the sections he presents as summaries. Francis Bacon never wrote about the drawbacks of the printing press. Cowen invented the whole thing, along with the references.

Having been called out on this, Cowen deleted the whole article without any explanation or acknowledgement, and appears to be currently pretending none of it ever happened. Only Google results (which still turn up his writing as a source) and the Internet Archive, which dutifully snap-shotted the piece, show that he ever tried it.

It’s also fairly clear that Cowen had help with his fabrication. Ask the latest versions of ChatGPT about Francis Bacon and you can generate similar nonsense-in-the-style of Bacon, complete with numbered summaries and invented references. One of Cowen’s invented quotes even appears to include one of his prompts pasted in without reading it - I’ve added bold and italics in the below:

He wrote: “But these three [inventions], perhaps, have fallen out by a certain fatality or providence of such a kind, that though they have added much to human power, they have not much increased human goodness; nay, rather, the first and last have furnished men with the means of doing more mischief, and the please say more second has made them more vain and arrogant.” (Bacon, 1605, Book I, Chapter I, section 5)

So, now we have another irony. The rise of large language models allow people to quickly generate plausible-sounding academic material and spread it around. It’s true that they’re not good quality at this point, but a) nor is much genuine academic writing, and b) the LLMs will only get better. And they can do this at scale.

It’s possible that people like Cowen - who appear to pride themselves on being able to put out vast quantities of writing on all topics both in print and online - see ChatGPT-like technologies as a way they can expand their reach further without doing all that tedious reading and thinking that so clogged up their lives before. This isn’t too much of a problem of course - how they generate their blog posts and their other academic-like-output-by-the-yard is up to them and their employers. If they think that churning it out every day in an automated fashion is fine, then they can continue being paid for it.

An instance of the side-effect

What is a problem is that Cowen / ChatGPT’s invented Bacon quotes have been dutifully indexed by Google and other search engines, and are presented in search results as genuine representations of Bacon’s work.

There will be more of these as time goes on, and as they get better and more convincing, they will get incorporated into genuine discussions. In this case, some online academics spotted a rat. But there’s never one rat.

As this continues, we will find that genuine knowledge - grounded in real facts rather than hallucinated ones, and genuine thinking rather than a plausible series of words - will become increasingly diluted with reams and reams of stuff that looks-like-it-but-is-very-hard-to-tell-the-difference.

Go back and read the (to emphasise, completely invented) Baconian worries about the printing press. The first in particular: the worry about the effects of flooding the market with useless books; this applies acutely to the process used to generate the fabricated worry.

But the fabricated Baconian solution: regulating the new technology and mandating that it be used better, does not seem a plausible solution; there seems no immediate prospect of getting LLMs to recognise the concept of “truth” (where do you find the concept of “truth” in vast vectors encoding the conditional probability of the next word?) And with so many people who believe that more words-with-their-name-attached-are-better, we must assume that these outpourings of all-too-plausible bullshit will continue. It’s more likely that we’ll just have to cope, and get used to it.

Novum Organum

We could forecast a rush to preserve the integrity of important writings. As the flow of bullshit increases, archivists, librarians and scholars will be given a new responsibility to preserve evidence chains back to original sources, and act as authorities to authenticate back to a pre-LLM time. One might even draw parallels to the treatment of hadiths, through trusted chains of transmission, rather than their inherent plausibility.

We could also forecast a rush to preserve real insight - genuine forms of mathematical and scientific argument rather than stuff-that-looks-like it. And of course the same technology can be used to generate unnervingly plausible statistics: safety data for medicines, mortality data for countries and events. Again, the reaction may to be raise the authority of expertise to some priestly class and ask them to preserve the integrity of real logic, reasoning and data, against this onslaught of plausible fakes.

Or, we might be lucky, and there may be a technological counterstrike. It could well be possible to develop tools that can quickly identify patterns in the work of LLMs, though on this, I imagine an arms-race familiar to us from spam filters and spammers will start and not easily end.

Honestly, I think we have to be humble, and acknowledge that we have no experience of a world where bullshit generation is quite so easy, nor so well rewarded. We should expect to be continuously surprised as things fall out as they may. New waves of bullshit are coming, and we as a society will cope with them as we can.

However, I can’t shake the idea that it would be helpful if people lodged in the highest academic positions were not quite so keen to act as agents to enthusiastically spread it around.

1. Cause of death - a chill caught when stuffing a chicken with snow, as part of an experiment to see if cold could be used to prolong life.

2. Also, one of the best ever stories of long-term misunderstanding, which - genuine or invented - makes me smile whenever I think about it.