What kind of rule has 171 adherents and 165 exceptions?
A phonics rule.
In their horrific, error-laden 2008 article, How Words Cast Their Spell, four ginormous phombie spelling experts (Moats, Joshi, Treiman, and Carrecker) penned the following:
"Instead of memorizing whether to use one or two consonants in the middle of words like cotton, tennis, sudden, muffin, and happen, students can use the rabbit rule. Of course, there are exceptions, such as cabin, robin, lemon, and camel, but these words are not as frequent as words that follow the rabbit rule."
I decided to put their assertion to the test. I searched online corpora for words with either six letters (rabbit, muffin, happen), or five letters (cabin, robin, camel), following the examples given, to see whether calling the latter set "exceptions" is accurate or fair or not.
I set loose parameters for the search results: I eliminated all words that had open-syllable prefixes like <re-> or <de->, and I eliminated all words with the suffixes <-s>, <-ed>, <-en>, and the comparative and agentive <-er> suffixes. I did not include words where a consonant doubles when adding a suffix, like robber or madden, and I eliminated compounds like hiccup. I eliminated words with a medial vowel digraph, like vowel or royal, because those don't really have medial consonants, and words with the stress on the second syllable and a reduced vowel th the first, like bidet or lapel. A handful of shorter words with final vowels snuck into the mix, like hello or limo, when their plurals showed up in the search. I used my own judgment to determine whether loanwords should be included. I kept some, like batik, and deleted others, like sisal, based on nothing more scientific than a general sense of how commonly they are used.
What I came up with was 165 words with one medial consonant (like habit), and 171 words with two medial consonants (like rabbit). That's really not much of a difference – certainly not enough to warrant just telling children to "use the rabbit rule." The search wasn't strictly scientific, but seeing the data in front of me revealed a lot more of the sense-making apparatus of English spelling.
A number of the double-consonant words can arguably be eliminated for having a base element that already has a double consonant in it, like cellar, mammal, ballad, billet, basset, or coffin, though their analyses may not be productive.
Just like with the consonant-<le> pattern, which I've written about before, there are common overarching patterns in which words take double consonants and which ones have single medial consonants. When we notice that certain letters either don't double or rarely double, then we understand the collection of words with single medial <v> or <x>, like vivid, divot, novel, civic, toxin, nexus, vixen. When we consider the particular phonology of the letters <g> and <c>, we understand why they are unitary in words like facet, tacit, logic, rigid. We notice that liquids <l> and <r> and nasals <m> and <n> are also common on the list of single middle consonants: color, filet, melon, salad, coral, harem, syrup, serif, comic, camel, timid, limit, venom, denim, panic, honor, and more. A single <s> can spell [z] as in visit, posit, rosin, risen, but a double <ss> cannot, as in fossil, lesson, vessel.
There are a handful of one-medial-consonant words that can be pronounced with either tense or lax vowels preceding them: basil, cumin, gala, satyr. There are some whose first vowel is not the default 'short' vowel for its spelling, but still falls under the lax vowel category: water, women, cover, sugar. Any of these is worth studying and has an orthography that is 100% explainable with regards to meaning, structure, and history.
As with any word study, it pays to look at the families. In the single-consonant families, we see the vowel pronunciation shift across the family: miser~miserable, panel~pane, manic~mania, pedal~bipedal, logic~logician, and so many more. Words with double medial consonants often have revealing single-consonant cousins: sorry~sore, hollow~hole, fodder~food, gutter~gout, butter~butyric, vellum~veal, mirror~admire, kennel~canine, and many more. Surely, in an article about etymology, for crying out loud, the authors might mention such a useful and productive understanding. You tell me, Dyslexics of Earth, which perspective helps more? Word families or the "rabbit rule" that barely works half the time?
Like many other resources, the 2008 article (which I've written about for more than a decade) is obsessed with so-called "short" and "long" vowels with regards to double or single consonants; they define the "rabbit rule" as "in a two-syllable word, there’s a double consonant in the middle after a short vowel." But again, this "rule" leaves us with pretty much the same number of "exceptions" as adherents.
Here's what's really going on: