Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

“Argument by analogy is very powerful and utterly fallacious” — my high-school Theory Of Knowledge teacher (Mr Dawson you’re the best)

That said, I’m really not enthused by this under either the business (relevant case history: Nest) or privacy perspective.



I agree with Mr. Dawson.

Argument by analogy is weak.

Communication containing analogies is powerful.

I do not find it useful to say "If A is like B, then all the things about B must apply to A." This forces us to find perfect analogies, and encourages too much bickering over how well the analogy applies.

But I do find it useful to use an imperfect analogy to illustrate something that you have already established directly. So establish that mergers between #2 and #3 rarely end up beating #1 directly, and then say "It's like..."

Our minds are not cold calculating machines. Colourful, humorous, and/or emotionally laden analogies do help us learn things and remember things.

I have found a similar thing going on with illustrations in blog posts. If you have a section about cascading failures in a digital service, and you include an illustration of dominos falling down, it does help people grasp and remember your point.

Even though obviously, cascading service failures are entirely unlike dominos, and it would be quixotic to attempt to reason about services from the things we know about dominos.


This is what I always try to tell people but I like the way you worded it: analogies are for communicative understanding, not argumentation.

All analogies are approximations. Even though they may not be argumentative, picking out flaws in analogies often presents as attacking a strawman.


If I can dynamically alter the spacing between the services ... or take a service out of the fall path before the crush comes. :)

Welcome back! I very much enjoy your meta cognitive viewpoint.


Analogies aren’t universally bad but they’re not really arguments.

You need to have an argument for the two things being in the same category for them to make sense.


Since I have drawn such unexpected interest in the subject and power of analogies (and the broader class of mappings they belong to, metaphors), might I suggest Surfaces & Essences by Douglas Hofstadter (of Godel, Escher, Bach fame) and Emmanuel Sander. It is a very insightful, powerful, and amusingly self-referential cinderblock-sized tome on how the ability to draw inferences, create and manipulate metaphors, and have a sense for where they do and don’t apply is a central issue for “general artificial intelligence” (you know, the MIT-cantered, Marvin Minsky-fuelled, LISP-encoded “symbolic AI” that eventually plunged into ‘winter’ only to emerge, phoenix-like but neutered, in our ‘modern’ statistics-based Machine/Deep Learning guises).

Analogies are certainly useful. I recall reading that Hawking first had the idea of his eponymous radiation ‘emanating’ from black holes by way of a calculation performed by a Russian physicist (whose name escapes me at the moment) that argued that a sphere of ferromagnetic material should ‘emit’ tiny but theoretically detectable amounts of electromagnetic radiation. I forget the details. But apparently Hawking’s brain drew the analogue between the inverse square laws at play, the virtual particles, and the non-permeability of the objects and set out on his quest for what eventually made him world-famous.

I myself am a prolific, if somewhat irreverent, manufacturer of analogies, metaphors, and euphemisms. I remember the general horror of my family members when I asked great-grandma why she had been moved to “the launch-pad ward”. In my defence, I was six.


A common antipattern here in HN comments is someone using an analogy (which is almost always bad), and then the responses devolving into arguing over how the analogy isn't correct.

I honestly think a "no analogies" rule for commenting would do more to facilitate good discussion than the existing "no jokes" rule. Analogies are almost never useful. Even the ancient Greeks realized this.


"You're painting analogies with an awfully broad brush."

----

"Painting with a broad brush" probably began as an analogy, but once enough people use it, it becomes an idiom. If understanding is shared between speaker/writer and listener/reader, idioms are a useful shorthand for sharing some commonly understood argument.

So useful, in fact, that people often forget what the original analogy was. For example, the verb "rewind." How many people using it think of reel-to-reels, cassettes, or videotapes?


"Painting with a broad brush" is not an idiom, since it's meaning is deducible from its content, not only its contex. It's a metaphor (literally false as written, but calls to mind an analogy with truth -- both the comment and broad brushes apply the same thing to many targets without any careful differentiation)

http://learnersdictionary.com/qa/Idioms-metaphors-similes-an....


Great link, thank you, I will use these words with more rigor in future. I think that “metaphor” is a much better word for the thing that is used to illustrate or add emotional oomph to an idea rather than to argue its merits.


Analogies can sometimes be a useful supplement to make an argument easier to digest, they just shouldn't be used as a substitute for a coherent argument. The latter tends to lead to what you're describing here:

>A common antipattern here in HN comments is someone using an analogy (which is almost always bad), and then the responses devolving into arguing over how the analogy isn't correct.


> Analogies are almost never useful

That statement is too broad to possibly be true.


And demonstrably false: what are numbers and arithmetic if not metaphors for acts performed on actual objects? “Get three apples, give away two, share what’s left with your sister” is encoded in the arithmetic analogy (3-2)/2.

Analogies... are more pernicious than you think.

Indeed, if one wishes to be extremely extreme, one might go as far as to argue that all language is ‘analogy’, insofar as it assigns symbolic monikers to real-world objects. Furthermore, phonetic encodings (alphabets) are a further level of analogy, as they posit that (for example) the “in” in “indeed” is “similar enough” to the “in” in “insofar” to be denoted with identical symbols. Those symbols are, themselves... you guessed it... analogies, metaphors, and other various kinds of abstractions.


Analogies are like the electricity of communication.


Well played.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: