In the decade leading up to 2020 we seem to have reached something akin to a tipping point with regards to the popular perceptions of the value of science. In times gone by, science held an extremely strong position in the minds of many non-scientists as scientific advances solved very important and very tangible problems that impacted many people's lives very directly. Because some of these advances were particularly easy to seem the institutions around science were very much trusted, despite the cutting edge of knowledge in the past being substantially behind where it is now.
More knowledge can lead to difficulties, a problem that isn't unique to science. This often happens when the results of increased knowledge reveal things that people don't want to hear. This has led to a number of attacks on the institutions of science when they convey these messages. For example, let's say you really like chocolate and some information comes to light that says only eating chocolate makes you sick from malnutrition, diabetes and other health conditions. You could react by saying "OK, I'll keep this in mind and weigh up how much I like doing what I'm doing versus the downsides," and then make a decision to keep doing what you were doing. Or you could reject the knowledge and just continue on doing what you were doing. When faced with inconvenience, it's the path of selectively rejecting knowledge out of hand that's increasingly been causing me concern lately. One of the most unfortunate developments has been the idea that there's no universal notion of truth. With a growing body of knowledge there's a growth in the number of inconvenient facts that you may have to deal with. Declaring a state of default on matters of truth forms an easy way out.
So there's a whole lot of things I've been writing about recently that touch on the topic of science. The personal tipping point for me was several years back when I realized that an organization I was working with that was ostensibly engaging in creating software for engineering and scientific applications had neither strong engineering fundamentals nor dedication to a scientific method. Their whole sales pitch was about applying mathematical optimization to increase the efficiencies of businesses, but internally they didn't practice what they preached in many ways. Upon talking with other people, it became apparent that this wasn't a rare state of affairs in the industry. In the intervening years I've rarely seen real data driven insights used by companies which compelled me to write about this. People are feeling the strain of dealing with an increased level of complexity across many industries, something that's particularly apparent in the modern tech industry. In the IT industry there has been a longstanding downwards trend in the average years of experience of those who are employed in the industry. This has had a bunch of effects, but it is only a small part of a far more complex story. While complexity has risen, there's not as much attention being given to managing that complexity. While we have seen a number of advances such as systems thinking and various other approaches, the rate of change has tended to outpace our abilities to deal with it. Again, I think this is partly because people don't want to hear inconvenient truths, and modern knowledge about management and systems thinking might not match the realities of an organization. I think a large contributor to this situation is that over the decades, science has enabled many advances that have drastically increased the complexity faced by people in that industry but taking on board the newer thought processes that have been discovered is far harder.
Part of what makes discussions about these topics difficult is that it seems like there's a lot of different things that people mean when they refer to "science". People sometimes talk past each other, even when engaging in discussions in good faith, which is very unfortunate. So to discuss a variety of trends in the industry, and the broader world, I realized I needed to take a step back and talk about what science actually is.
How did we get here: some history of science
A long time ago, I ended up taking some classes about the History and philosophy of science. Perhaps one of the most striking things to me at the time was that most of the people taking this course weren't themselves practicing or studying science (perhaps including me since I was studying mathematics mostly1). Such a state of affairs was not uncommon, and many scientists from the era where science was at a peak were openly disparaging about the topics of the history and philosophy of science:
Philosophy of science is about as useful to scientists as ornithology is to birds.
This quote is attributed to Richard Feynman, a very famous scientist, and I think it summarizes an attitude that was widely held in the scientific community many decades ago. I can't help but wonder how his experiences with NASA working on the Rogers commission in the aftermath of the Challenger disaster, something he writes about in the book "Why Do You Care What Other People Think?", would have impacted this position. Specifically because the institutions of science are far more answerable to the opinions of the general populace than they were before, what people think of science matters far more for the scientific institutions than it previously did.
Perhaps one of the biggest changes in the last 30 or so years has been the huge effort put into the undermining of scientific institutions. The same power from knowledge that led to things like the development of the atomic bomb and various other world-changing inventions can shift the balance of power irreversibly. And it's exactly this power that can threaten the interests of various groups. The power of empires is built mostly upon the ability to exert force on others, and scientists — particularly physicists — had an alignment of interests at certain points in history, especially in the era of industrialized warfare when scientific advances allowed more projection of force. Not to mention instances of total war where scientific advances allow more industrial output to fuel the war machine.
But what happens when the cutting edge knowledge found by the application of the scientific method starts to clash with existing interests and other structures of power? A good example of this can be found with the extreme disinformation campaigns from big tobacco companies. As late as 1994, deliberate lies were being made by Big tobacco executives about the health of their products. The pressure of the truth of the matter just kept building and building over time. Regulations were eventually adopted in many countries, forcing the sale of such products to come with health warnings. A big tipping point came in November 1998 when the tobacco industry reached a deal known as the Tobacco Master Settlement Agreement which exempted the tobacco industry from legal liability for the harm caused by tobacco consumption in return for a number of restrictions on advertising and annual payments to the US states. This settlement was a matter of legal and political exigency where the $200 billion in guaranteed payments was likely the cheaper option. This was mostly a disaster because the states issued bonds with unfavorable terms which mostly destroyed the financial value of the deal2, finances that were supposed to fund anti-smoking campaigns. Luckily, despite all this, the consensus that tobacco is harmful for you has strengthened since the 1990s.
Fast forward a couple of decades and the consensus is now very strongly that tobacco smoking is harmful, to the point where attempting to suggest otherwise is seen as an expression of aggressive ignorance. Facing this change, some of these same companies have capitulated on their earlier position by publicly admitting what they always knew which is that smoking is harmful to your health. The admission is part of a marketing pivot, this is to suggest that e-cigarettes are actually far better for your health than inhaling a bunch of toxic smoke byproducts. The fact that they even admitted their old flagship product is unhealthy — the same product they invested so heavily in protecting by expending massive effort in lying to the public about — is a huge turnaround, but such statements don't represent a sudden change of approach to discovering the truth. These statements are being made because of economic motives (a theme that comes up repeatedly when dealing with this topic), and with a substitute good that's vastly better (e-cigarettes), there's now an option to shift to a different product. These aren't just carefully crafted PR words either, Big tobacco is buying up interests in e-cigarette companies, the pivot to e-cigarettes is a carefully considered move to retain the commercial empires they fought so hard to build. Some of the statements from Big tobacco that are disguised as comments about knowledge are not actually seeking knowledge at all and this is a distinction that's absolutely critical to make. The point of all this isn't to pile on to certain companies but to highlight that statements that are ostensibly about matters of fact aren't necessarily made with a goal of finding the facts of a matter, and can actually be carefully crafted to bury the truth.
One thing that's especially important to note is that deliberate misrepresentations of the truth is not even the biggest threat to scientific progress. When someone is lying, there's an acknowledgement of what the truth is and an effort made to misrepresent it for gain. The rising cultural trend of bullshiting, where the matter of the truth isn't considered at all is far more pernicious. The bullshitter doesn't care about the truthfulness of their statements at all because they aren't searching for knowledge in their discourse. The bullshitter talks only as a means to gain power with absolutely no regard for the truthfulness, or lack thereof, of their words. In terms of repercussions for the search for knowledge and truth, bullshitting might be more dangerous than lying.
The collateral damage to the prestige of the scientific method from all these brutal disinformation and propaganda campaigns — along with the rise of bullshitting culture — is an externality which we collectively are likely to pay for, perhaps heavily, as recent world events are showing. A decline in good faith efforts to find the truth, of which the scientific method is just one approach, is something I find deeply disturbing. The doubts about the scientific method that were sowed to prop up the greed of certain individuals and companies has unfortunately created a generation with confusion about the value of truth-seeking activities and a decreased prestige for those who are engaged in those occupations. This has made discussions in the general public on the topics of science harder to have. The cost of abandoning efforts to find the truth is hard to quantify exactly but appears to be enormous.
We would be very wise to carefully distinguish statements that have their power derived from knowledge that allows predictive power and construction of wealth against those that are backed by the power of the barrel of a gun. So how do we derive knowledge that is powerful, in isolation of the societal structures that we have created? Is there knowledge that is discovered instead of created? How can we independently verify the truth of a statement? Perhaps the biggest goal of science is to create an unbiased view of the world that is based on evidence and careful reasoning from that evidence. A method with which we can build up a body of knowledge in this way is the scientific method.
I remember a recommended introductory book on this topic from the course I took at university, called "What is this thing called science" by Chalmers; and a far more thorough book which was an edited collection of papers called "Philosophy of science". I still have this book on my shelf and thinking about these topics has been cause to read it again.
Both books cover an important distinction that must be made between the practice of using the scientific method and the communities and organizational structures that have come to exist around people practicing things called science.
I will talk about the institutions of science in later posts but first a good discussion of the scientific method is needed.
The practice of science
When we talk about people who are working on gathering knowledge in a scientific manner, we are referring to people who are using the scientific method to gain insights.
The scientific method is a process that when followed allows us to gain insights in a certain way. It's a way in which we can discern true and false statements about the world.
The power of science is mostly vested in the strength of the scientific process to enable us to gain insights into how the world works.
The basics - falsifiability
Being able to explain why things happen is incredibly important. We refer to a proposed explanation for why something happens as a hypothesis.
One of the most important things about scientific statements is that they can be shown to be false. This property is known as falsifiability. This is especially important because if you cannot show something to be false you lose the ability to claim that it is true.
A scientific claim must be falsifiable.
Prefer direct evidence
The scientific method is built around the interaction of claims made by hypotheses and the evidence that backs those claims up. The processes of the scientific method gives us a way to verify or falsify claims via experiments. In this way science is fundamentally about not needing to trust experts, anyone who follows the process should be able to verify any scientific claims made via their own interactions with direct evidence. This is not to say that such a mindset stipulates you should reject anything coming from experts but rather that scientific knowledge must be built up in a way that requires no appeal to expertise.
A fundamental part of this idea is that you should be able to verify any scientific claim via the ability to conduct an independent experiment. The ability to independently verify a scientific claim is one of the most fundamental parts of science, without this you don't have a scientific claim.
Experiments have to be able to be verified by evidence to be valid.
This is why we have traditions such as peer review, when an experiment is verified independently of the original authors it makes a substantially more powerful statement. Any author can make mistakes or in the case of say big tobacco in the early days be corrupt and bought out. Multiple independent verifications of an experiment that are carried out from first principles makes a strong statement about the hypothesis being tested. With enough such evidence people may start to postulate theories that are based on the evidence observed.
The notion of a scientific theory warrants a lot of further discussion since in practice this has become a key battleground for people claiming to have truthful claims.
Being able to replicate an experiment is an important part of the scientific method. This is something that's a particular problem in the modern world which has faced an extreme replication crisis in recent times. I'm planning on writing an article about replication of scientific work that uses computer software as a component and perhaps to talk on why the replication crisis has become so severe. Much of this crisis has come about because of systematic factors related to the incentive structures in the institutions around science, something I'll talk about in other articles.
This is slightly different from the issues around building from evidence as there's a lot of methodological and technical reasons why an experiment that deliberately works with direct evidence and data may not be able to be replicated.
Independent replication is an important guard against errors and personal biases. Replication also makes a case for the predictive power of a statement, science derives much of its practical power for this.
An experiment that cannot be replicated is a poor experiment.
If you can't replicate an experiment you can't reliably build on top of the claims made.
Mathematics has a different feel to it than many scientific domains because deriving knowledge from empirical evidence plays less of a role in mathematics compared to the natural sciences. Mathematics tends to gather some set of axioms then build from those in a very directly verifiable manner. Which axioms are chosen is a very interesting question, since not all sets of axioms are of equal use. That said there's an uncanny overlap between mathematics and the natural sciences that was commented on in the 1960 paper The Unreasonable Effectiveness of Mathematics in the Natural Sciences, though this might be just because a sufficiently refined understanding of phenomena can often be represented in mathematical (ie. exact) terms. ↩
Many of the states that were party to this agreement wanted faster access to the money that the tobacco companies had agreed to pay over the years. To try to get the money soon after the agreement in one big payment rather than wait they turned these into financial instruments via the bond market. The catch, these bonds which were backed by a legal settlement with the tobacco industry were seen to be very high risk, so the bonds had to be sold at a small fraction of the price in order to get anyone to invest in them. The result being that much of the payments from the big tobacco firms didn't go to the states who made the settlement but rather ended up in the hands of the bondholders. Instead of paying yearly interest the bonds were set up as capital appreciation bonds, meaning that all the interest is repaid at the maturity -- which is conveniently long after any of the legislators who agreed to the terms are gone. Moral hazard at it's finest. But it gets even better because there's a long time to maturity and the whole reason the bonds exist in the first place is because of a settlement that directly aims to reduce the power of the companies that would be paying back the bonds. Creating a long term bond with the express purpose of using the money from the bond to advertise a campaign to make the position of the companies paying the bonds back worse... I mean what could possibly go wrong? This whole situation is just a policy and economics dumpster fire. ↩