back
70 comments
Interesting language analysis, but I disagree with the conclusion. Authors merely stating "conclude" or "determine" instead "believe" or "feel" doesn't prove they actually were any more or less rational or logical. The latter use could be more truthful to ones internal thought process.

[If I write "I believe <blah blah>" because I believe something and state the reasons why I believe something, it is probably more informative than the output produced if I believe <blah blah>, then search Google for some random authoritative looking sources that say something that superficially matches what I want to say, and then write "these authoritative sources conclude <blah blah>"].

The observation indicates that popularity of some turn of phrase and "key words" over another have changed over time, which is interesting. But for determining how the actual "rationality" of content has changed, one should pick some set of random articles from different periods and actually analyze structure of arguments presented.

Yeah, maybe the outward pretense of rationality was more important in decades past.

In my experience reading scientific papers, "we conclude" is just a pretentious old-timey way to say "I think". It never actually meant anything more rigorous, it was just the norm to write more pretentiously and inflate the certainty of conclusions.

We've been greatly humbled by being wrong over and over in recent decades. Maybe that has driven people to be less confident and stop presenting their thoughts as inevitable rational conclusions.

Another illustration of the weakness of their data interpretation -- the figure shows use of the word "data" has declined in the past 20 years. Has actual use of data declined over this period? Quite the opposite, surely.

I think (haha) this last point is actually quite revealing and strongly supports the "decreasing pretension" interpretation of the paper's word use trend charts. Maybe people stopped saying "the data says" over and over in a crude attempt to box the reader into submission.

It's hard for me to appreciate an argument that we were more rational during a time of religious dominance, the last 2000 or so years.

We've been greatly humbled by being wrong over and over in recent decades. Maybe that has driven people to be less confident and stop presenting their thoughts as inevitable rational conclusions.

This resonates with me. We are more able to verify our assertions, quicker, faster, more accurately and are more hesitant to make bold claims.

I also think we have a bit of division of specialization going on, where the world keeps turning if everyone is competent in their specialty, allowing our personal, unprofessional, side to act more irrationally without as much consequence as in the past.

Ultimately it's really hard for me to believe that as we spread education and opportunity to the 8 billion people of the world, we are somehow measurably less rational then when a large swath of our population couldn't even read or write.

> In my experience reading scientific papers, "we conclude" is just a pretentious old-timey way to say "I think". It never actually meant anything more rigorous, it was just the norm to write more pretentiously and inflate the certainty of conclusions.

A person of a certain age remembers specifically being taught to obscure weak I-statements with impersonal statements that one could conclude are more rational if they do not think about them too hard.

> In my experience reading scientific papers, "we conclude" is just a pretentious old-timey way to say "I think".

My high school Philosophy teacher would argue that "I think" is a weasily way to avoid accountability and challenge. And, "I think", she would say, is implied by any first person statement.

Nobody can challenge what you think; your thoughts are your own. What they can challenge are declarations you make about the external world. What she failed to foresee is the degree of offense people would eventually claim regarding others' declarations. "I think" is indeed a way to pull your punches, but often to avoid conflict with the public's increasingly hyper-sensitive disposition than to protect one's own fragile ego.

The shift in language in culture as a whole has (in my own personal observation) been directly tied to the inability of the general populace to even understand basic logic. It also corresponds to a direct rejection in much of academia that there is such a thing as Truth.

If you "personally believe" something in today's context, then it is "true for you" and cannot be questioned.

Also, I'll just leave this here for fun: https://www.youtube.com/watch?v=lj3iNxZ8Dww

Hypothesis: We have more books authored per capita today than then, suggesting a broadening of the types and training of authors.

"In 1907 there were 9,260 books published in the United States according to the New York State Library Bookboard. ...Compare that to 2010 when there were 316,480 books published by traditional publishing companies according to Bowker." [1]

Meanwhile the U.S. population has grown from 92,228,496 in 1910 to 308,745,538 in 2010. [2]

Calculate ratios. In 1910 there was a book published per every 9,959.0 Americans. In 2010, there was a book published per every 975.5 Americans. Ratio of ratios: 10.2x in 1910 individuals per book published.

I'm not going to bother with any stat test. There has been a magnitude of change during this span. We have more authors (or publications) per capita than before, likely reflecting a broadening of the socioeconomic and educational classes of the authors.

[1] http://stuffnobodycaresabout.com/2012/01/31/how-many-books-w... [2] https://en.wikipedia.org/wiki/Demographic_history_of_the_Uni...

EDIT: It seems I should have started from 1980, not 1910...but now its time for work.

more stats about bbbooks: https://ourworldindata.org/books
I believe you are right!
> Interesting language analysis, but I disagree with the conclusion. Authors merely stating "conclude" or "determine" instead "believe" or "feel" doesn't prove they actually were any more or less rational or logical.

It actually does. The primary difference is first-person pronoun use versus reliance upon external factors, such as evidence of any kind. That shift indicates a trend towards narcissistic thinking and possibly isolation.

This is ridiculous. "We conclude" means you relied on evidence, "I think" doesn't? Give us a break.

If you have any experience reading scientific papers, you should know that this sort of language carries no information whatsoever on the rigor of the author's methodology.

I see the current trend towards first person, and away from passive voice, as the opposite. "I preprocessed the data using algorithm X", compared to "The data was preprocessed using algorithm X", explicitly acknowledges a fallible human being at the centre of the analysis.
It could also reflect a trend towards more precise use of language and more accurate communication of the outcomes of the author's work. You'd have to do a more in depth study of the actual work and usage to determine which is the case.
According to the article rationality started to decline at ~2000, and about at the same time psychology came short of a rejection of all the prior research due to deficiencies in a methodology. Maybe it is not a decline of a rationality, but an better understanding of a real value of a scientific result?

Also at 2001 Judea Pearl published "Causality: Models, Reasoning and Inference", which gives mathematical tools to track causes and effects. This book is full of "beliefs". Less so then the previous book of the same author ("Probabilistic Reasoning in Intelligent Systems"), but nevertheless.

I believe, that the use of word "believe" or "feel" is not a sign of rejecting rationality. In this case it is about accepting the idea that we do not know the absolute truth and inviting others to challenge our beliefs. It fits nicely with ideas of Pearl's causality which states that it is impossible to reason about causes while sticking strictly to objectivity. So I feel it differently: it is not people became less rational, it is the rationality itself became less objective.

> According to the article rationality started to decline at ~2000

From the article:

What precisely caused the observed reversal of the long-term trend around 1980 remains perhaps even more difficult to pinpoint.

Eyeballing the chart, "rationality words" peaked somewhere there.

Yes, you are right. I concentrated on another graph, that shows the use of "intuition" words starts to grow closer to 2000. But it fits perfectly to a narrative of "The Book of Why" of Judea Pearl. He tells the history of statistics, and around 1980 there was growing frustration with logic and statistics. AI researchers was unable to build AI on a basis of formal logic (pattern matching, expert systems). So various probabilistic logics were spawned (fuzzy logic, bayesian analysis) they were trying to merge formal logic and statistics into a single framework. But it was not enough because it didn't allow to speak about causes and effect. And formal tools to deal with causality were developed in 1990s. Exactly when the "rational" word use declined and the "intuition" word use grown.

We see in graphs how people stopped using "rational" words (because they saw that they didn't work?), and closer to 2000 they've learned a new vocabulary. Correlation between the history of statistics and this graphs is just perfect. Though it is just a correlation and moreover it is a hindsight correlation, not a good reason to be overexcited and to propose a causal explanation.

1979 is the year Baudrillard published Simulacra & Simulation (a sort of obsession for me) and became the philosophical superstar of postmodernism, documenting a great cultural sea change towards a more cynical, superficial, often delusional worldview that pervaded the 1980's and echoes today. He was more inflammatory than objective/rational himself, however.
>In this case it is about accepting the idea that we do not know the absolute truth and inviting others to challenge our beliefs.

Empirically, does it seem we're in more open times where we're more "inviting others to challenge our beliefs"?

I suspect the issue is the separation of academia and the mainstream discourse.

In academia, these days, you can probably question anything and everything - revisionism is rampant in all sorts of areas.

In the mainstream, though, that is not necessarily the case.

It's a pretty bold leap to find a shift in language and then presume that this represents an underlying change in how people reason.

It would be equally believable that people are simply using different, more fashionable, words, but reasoning the same as they have always done.

I've been reading and criticizing media-hyped studies for a long time, and I've noticed there is an "interestingness bias" in these studies' conclusions.

Media-hound scientists will always walk straight past the less interesting explanations of their data and seize on the most attention-grabbing conclusion as the one "the data clearly suggests."

Then the comment section will go "what about the less interesting explanations"? Which of course were downplayed in the paper, and omitted entirely from the media promotion of the paper.

It's utter bollocks.

And it even looks as if they've assigned "intuition" vs "rational" words afterwards. How come "truth" is intuition while "determine" isn't? But adding "facebook" is really the nail in the coffin: that only became available in 2004 or so, and so could almost only show an upward trend.

I think it is not unreasonable to believe language (and word use) and thinking patterns are connected. The fact that these relate (and yes, that different words lead to different thinking patterns) is well understood in practicing areas of psychology.
I remember being taught (or reading somewhere) that using expressions like "I believe" or "I feel", rather than "I think", can be more persuasive, because they're more open-ended and less assertive (which can put people off). Whereas something dispassionate like "we conclude" or "the evidence shows" would be more appropriate for academic discourse.
I feel this is true.

Often conversations evolve from describing a first person experience "I went to the shops" to "you know, when you get there and XYZ.. and how you have to..." I feel a lot less receptive when this shift happens.

Another example "you should try fasting" instead of "fasting felt helpful for me".

I find “believe” and “feel” to be more natural in relation to a fact which is still under investigation. As more affirmative samples come in, I feel and believe more strongly in its truth.

“Conclude” implies conclusion, where no further evidence is needed, and where no evidence to the contrary is expected. The history of science suggests the vast majority of results are not conclusive.

So maybe this language is not entirely a measure of rational vs emotional, but also a measure of confidence.

Nevertheless, I think the conclusion of this article probably is correct.

I'm not sure I would take their work at face-value. These studies of word frequencies over time are often too atomized in their failure to consider the relationships between words at all.

And of course, such a change in speech patterns doesn't have to actually speak to a decline in rationality but to a stronger focus on the author of a claim. The rational process of thinking might be as prevalent nowadays, it's just differently worded.

The idea that the meaning of words changes along with the frequency of their usage seems pretty important to linguistic studies.

For example, it's not hard to imagine data like this being used to say that the there is "less cultural interest in political correctness due to it's declining usage in discourse" while not factoring in that "woke" has largely replaced it in the frequency of usage while often representing the same concept in the speaker's mind.

The way language and speech change is extremely complex, and I feel like an analysis that uses that time span starts to become almost worthless as anything other than a trivia fact. "People speak differently than they did 20, 50, or 100 years ago" was already pretty well understood, and the assumption that these words have inherent, immutable meaning is misrepresenting what words actually are, stand-ins and generalizations limited by the speaker's understanding of them and the audience they're spoken to.

Saying "conclude" instead of "believe" does not make your conclusion more logical or less emotions based. Many irrational and emotion based opinions are masked by sure language.
It might be linked to the rise of the internet where it became easy, on any topic, to find data that directly contradicts any other data. There was a lot less data available in 1990. Anyone who makes a habit of saying "the data shows..." is going to experience a lot of bruising episodes where they learn how wrong they were.
It's not fair to equate terms like 'data', 'analysis' and 'result' with 'reason'. In fact, terms like 'think' and 'believe' are more rational because they leave room for doubt - This aligns better with reality since we can never be 100% sure of anything.

Statistics can be manipulated in many ways; data could be collected incorrectly (introducing bias), could be aggregated and combined incorrectly and then searched and tabulated incorrectly. To assume that 'data' is always correct and paints a clear picture shows a lack of rationality.

Ironically, while this is a very data-driven piece of research, the fact that it chimes with how most people (especially within the HN crowd) 'feel' will make it more persuasive!

I hope that some of this trend is due to the acceptance thresholds lowering slightly, allowing for publication of incomplete research, on which others can build, rather than requiring a complete conclusion. Of course, one may draw conclusions about funding causing people to wrap up research prematurely.

> Ironically, while this is a very data-driven piece of research, the fact that it chimes with how most people (especially within the HN crowd) 'feel' will make it more persuasive!

Well said! I wonder if the crowd here will realize they're only seeing this article because so many of their compatriots felt good enough about the article to upvote it.

"I conclude" doesn't solve the problem either. It's still an attempt to make a conclusive statement in the midst of the never-ending epistemological crisis.

If we are to remain rational observers of reality, it is far better IMO to frame statements as "thus far, our hypotheses have / have not been invalidated based on the following experiments, which can be reproduced by following these exact steps..." along with proper control groups, etc.

Or maybe we just use the same words to convey different things over time.

I often find myself saying "I feel like <some argument> might be the case" when it would have been just as accurate to say "I think <some argument> might be true"

I don't think it makes me any less rational to use the word "feel" over "think", what makes me more or less rational will be the content of <some argument>

I believe I am lost in translation, but I could also have concluded as such I guess.

This is even less meaningful than word clouds in my opinion (final conclusion pending).

I'm not sure on the actuality of reduction in rationality but using shifting meaning and changes in phrases just strikes me just a slight step up from reading tea leafs.

Yes if there was some obvious shift of everyone referencing religious texts or astrological forecasts that would be one thing. Going from I concur to I believe is just a stretch to state anything about rationality.

Aternate hypothesis: the rise of more subjective terms parallels increased social justice movements, and thus reflects the attempt to soften language to appease those who might vehemently disagree with “conclusions” that you might draw.

Thus, it is not the authors have declined in their own rationality, but are instead increasingly choosing to rationally avoid the increasing levels of conflict now present in every aspect of discourse.

Certainly, that explains why I avoid making statements without couching them in such doubt, even when I rationally know that such diminishes those statements potential impact. I’m better off falling short than getting knocked down.

The article is relatively ok though I disagree that not being rational can be attached to individualism. There is one dentece suggesting that. Looks like politics shoehorned into the article.

My opinion is that rationalism can only come from individuals.

Only individuals exist. There is no such thing as a collective brain.

Similar themes are echoed in Steven Pinker's 'Rationality' or 'Enlightenment Now'.
> The authors did find that the shift from rationality to sentiment in book language accelerated around 2007 with the rise of social media, when across languages the frequency of fact-related words dropped while emotion-laden language surged, a trend paralleled by a shift from collectivistic to individualistic language.

Engagement algorithms, most likely. People are probably more likely to engage longer if they're emotionally stimulated than bombarded with a bunch of pesky facts.

I believe that we conclude... why have one when we can have both?
Link to actual paper: https://www.pnas.org/content/pnas/118/51/e2107848118.full.pd...

Link to corpus (it's paywalled): https://www.science.org/doi/10.1126/science.1199644

Link to the democratization of fiction publishing: https://justine.lol/fiction.png

I see this most directly occur on HN in discussions about housing prices. People feel threatened when challenged by facts, data, or the lack thereof.
Sad and true. Not funny.
We are taught not to assail someone's feelings.

It's much easier to criticize someone's conclusions.

At least, that what I feel.

Empiricism has context. It was weaponised by a crackpot senator in Queensland who doesn't understand science and used as a cudgel to fuel his rants. His point appears to be that nothing he can't understand can be true, and since elected (upper house elected, party ticket elected, not actually individually elected) he is the arbiter of truth.
Sounds like Trudeau... Is this senator 100% wrong all the time? Which bits are wrong?

Sounds more like you've just used this comment section to launch your own rant against a voice that's too loud for your liking...