Home
About How to Use
Taxonomy Glossary About the Author

The History of the Taxonomy

Beginning with Aristotle, the first logician to name fallacies, most logicians who have studied fallacies have classified them into types. Aristotle classified his list of fallacies into two types1:

  1. Linguistic: Those that depend on language.
  2. Non-linguistic: Those that do not depend on language.

Subsequent logicians have usually extended Aristotle's classification by subdividing the second, non-linguistic, category into sub-categories, such as fallacies of relevance, of presumption, and of ambiguity2. However, most such classifications have remained relatively "flat", with all fallacies on the same level. Unfortunately, a flat classification does not do justice to the complexity of the logical relations between different fallacies.

The Fallacy Files Taxonomy is a tree-like structure that classifies all of the fallacies in these files by the sub-fallacy relation. A sub-fallacy is a specific version of a more general fallacy that has whatever features the more general fallacy has, together with specific features that set it apart and make it worth naming in its own right. For example, in the Taxonomy, instead of grouping together "fallacies of relevance", there is one most general such fallacy―namely, Red Herring―and all fallacies of relevance are sub-fallacies of it. Red Herring is itself a sub-fallacy of Informal Fallacy, which is a sub-fallacy of the most general logical fallacy of all: Logical Fallacy. Logical Fallacy is, thus, the top node of the Taxonomy, for every fallacy in the Taxonomy is a sub-fallacy of it. The sub-fallacy relationship is like a tree with a trunk―Logical Fallacy―which branches until it reaches leaves, that is, fallacies that have no sub-fallacies―for example, Appeal to Celebrity.

There are three versions of the Taxonomy:

How to Use the Taxonomy

The Taxonomy is more useful than the alphabetical index for studying the logical relationships between fallacies. To understand an individual fallacy, it may be helpful to move upward in the Taxonomy―or to the left in Version 1.0―in order to understand the more general fallacy of which it is a sub-fallacy. Also, moving downward―or to the right in Version 1.0T―can help in understanding a general fallacy through seeing more specific versions of it.

In addition, fallacies that are sub-fallacies of the same general fallacy are like siblings, since they share the same parent. So, it may help to compare and contrast a fallacy with its siblings. As with human siblings, the likeness between sibling fallacies is stronger in some cases than in others. For instance, the causal fallacies Post Hoc and Cum Hoc are more similar to each other than they are to other siblings, such as the Regression and Texas Sharpshooter fallacies. In Version 1.0, this strong sibling relationship is indicated by a thicker, similarly-colored line connecting the two fallacies. In the later versions, unfortunately, this close sibling bond is not explicitly represented, but similar fallacies are next to each other on their common branch.

Another use for the Taxonomy is in finding a fallacy whose name you don't know, but you do know what general type of mistake you are looking for. Start with a general fallacy, and drill down into the Taxonomy―that is, moving down in Version 2.0 and 3.0 or to the right in Version 1.0―until you find what you're looking for. Happy fallacy hunting!


Technical Appendix: Mathematically speaking, the sub-fallacy relation is a partial ordering of the set of fallacies. In terms of graph theory, the Taxonomy is a cyclic graph. As mentioned in the section above on Version 1.0, it is not actually a tree in structure because some fallacies are leaves on more than one branch, though this characteristic cannot be shown in the more recent versions for technical reasons. Specifically, at the level of HTML, Version 2.0 and 3.0 are simply nested lists. While it would be possible for the same fallacy to occur in different sub-lists, this would create duplicate nodes rather than show a single node on multiple branches. The taxonomies are already tangled enough without this additional complication.

Version 1.0 is an image and uses an image map to create the links from the nodes to the individual fallacy entries. As you can imagine, this made it extremely difficult to update it with new fallacies since the image would need to be redrawn and much of the image map changed. As a result, Version 1.0 is missing many fallacies listed in later versions, such as Overgeneralization, the Lottery Fallacy, and the Fallacy of the Sheep.

As mentioned above, both versions 2.0 and 3.0 are nested HTML lists and the tree diagram is drawn using Cascading Style Sheets―that's right, no Javascript needed! This was the main reason for adopting the new versions as they are much easier to update than the original. Unfortunately, some of the structure in the old version cannot be rendered in the new ones, namely, the close sibling relations and the graph cycles.

Also, three fallacies have been temporarily omitted from versions 2.0 and 3.0: Composition, Division, and Loaded Question. The latter does not fit well in the taxonomy as fallacies of erotetic logic―the logic of questions and answers―may require their own taxonomy. Composition and Division are awaiting a more general super-fallacy that currently lacks an entry. The Taxonomy, as the rest of The Fallacy Files, is a work in progress.


Notes:

  1. Aristotle, On Sophistical Refutations, Translated by W. A. Pickard-Cambridge, Part 4.
  2. This is the classification scheme of the popular textbook by Irving M. Copi & Carl Cohen, Introduction to Logic (11th edition, 2001), chapter 4.

Acknowledgments: Thanks to Daniel Nussdorf for pointing out a mistake in Version 2.0 and to Nicholas Joll for some questions and criticism of 2.0 that led me to construct 3.0.


Revised: 9/4/2025