2.44 Chance

Category: Philosophy of Science

Keywords: chances, chance, probability, probabilities, prob, frequency, random, probable, frequencies, sample, coin, probabilistic, propensity, bayesian, distributions

Number of Articles: 507
Percentage of Total: 1.6%
Rank: 12th

Weighted Number of Articles: 398.2
Percentage of Total: 1.2%
Rank: 25th

Mean Publication Year: 1977
Weighted Mean Publication Year: 1979.4
Median Publication Year: 1978
Modal Publication Year: 2007

Topic with Most Overlap: Theory Testing (0.053)
Topic this Overlaps Most With: Theory Testing (0.0731)
Topic with Least Overlap: Emotions (0.00027)
Topic this Overlaps Least With: Belief Ascriptions (0.00039)

A scatterplot showing which proportion of articles each year are in the chancetopic. The x-axis shows the year, the y-axis measures the proportion of articles each year in this topic. There is one dot per year. The highest value is in 1945 when 2.5% of articles were in this topic. The lowest value is in 1901 when 0.0% of articles were in this topic. The full table that provides the data for this graph is available in Table A.44 in Appendix A.

Figure 2.106: Chance.

A set of twelve scatterplots showing the proportion of articles in each journal in each year that are in the Chancetopic. There is one scatterplot for each of the twelve journals that are the focus of this book. In each scatterplot, the x-axis is the year, and the y-axis is the proportion of articles in that year in that journal in this topic. Here are the average values for each of the twelve scatterplots - these tell you on average how much of the journal is dedicated to this topic. Mind - 0.9%. Proceedings of the Aristotelian Society - 0.4%. Ethics - 0.2%. Philosophical Review - 0.5%. Analysis - 1.2%. Philosophy and Public Affairs - 0.2%. Journal of Philosophy - 1.0%. Philosophy and Phenomenological Research - 0.6%. Philosophy of Science - 3.3%. Noûs - 1.1%. The Philosophical Quarterly - 0.6%. British Journal for the Philosophy of Science - 5.0%. The topic reaches its zenith in year 1947 when it makes up, on average across the journals, 2.3% of the articles. And it hits a minimum in year 1901 when it makes up, on average across the journals, 0.0% of the articles.

Figure 2.107: Chance articles in each journal.

Table 2.93: Characteristic articles of the chance topic.
Table 2.94: Highly cited articles in the chance topic.


I’ve called this chance, and it does have a lot to do with chance, but a more perspicuous name would be philosophers of science talk about probability because the big difference between this and formal epistemology is that the folks here are philosophers of science, and the folks there are epistemologists. There are a lot of papers about chance in here—the second characteristic article is Michael Strevens responding to David Lewis’s revision of the “Principal Principle”—but there are also a lot of papers about probabilistic reasoning in scientific methodology.

Now, I was very surprised that the model could tell the difference between the philosophy of science articles and the epistemology articles. I have a reasonably strong “know-it-when-I-see-it” sense of how philosophy of probability gets divided up into philosophy of science and epistemology. But I wasn’t sure it was anything more than a set of gut reactions/prejudices, and I would not have expected at all that a machine-learning algorithm could replicate it. Yet here we are—the model really did do a very good job of splitting them up.

It’s not like it doesn’t think the two are connected. This topic and formal epistemology are two of the closest connected topics. But it does just enough to tell them apart. We can see this by looking at two David Lewis articles.

Table 2.95: David Lewis, “Humean Supervenience Debugged.”
Subject Probability
Chance 0.3773
Ordinary language 0.1908
Composition and constitution 0.0641
Time 0.0627
Laws 0.0616
Modality 0.0488
Thermodynamics 0.0370
Formal epistemology 0.0231
Table 2.96: David Lewis, “Sleeping Beauty: Reply To Elga.”
Subject Probability
Formal epistemology 0.7362
Arguments 0.0639
Chance 0.0599
Theory testing 0.0452
Modality 0.0421
Time 0.0352

That’s pretty good; in each case the model gives the intended topic 10 times the probability it gives the unintended topic. It’s not perfect; I really don’t know why it thinks the Sleeping Beauty paper is just about as likely to be a Modality article as the Humean Supervenience article. But this is a quibble; I’m really happy that the model found this distinction. It made looking at the splits between Epistemology and Philosophy of Science much easier.