I before E, except after C

I want to work through a very simple illustration of some surrogation mechanics, to hopefully lay groundwork for future thought:

There’s an old spelling rule from grammar school, “i before e, except after c.” It’s a heuristic, or rule of thumb—it isn’t perfect, but it’s helpful, for words like “tier” or “receipt.” Complete knowledge of spelling rules—that is, of every English word that uses “i” and “e” sequentially—requires a lot of information (i.e. specification). Our heuristic lossily compresses that full specification into a much shorter description, albeit one that would mislead a naive follower, should he attempt to evenly apply it. (For instance, in the word feign, or the name “Stein.”)

Let’s try to catalogue and re-compress the full set. Exceptions will be (1) “e”s that precede “i” despite being preceded themselves by “not-c”—or else (2) “i”s that come before “e”s despite being preceded by “c”s.

Exception 1 (not-c/e/i):

• Caffeine
• Codeine
• Eight
• Feign
• Forfeit
• Foreign
• Forfeit
• Freight
• Heifer
• Height
• Leisure
• Protein
• Neighbor
• Neither
• Seize
• Surfeit
• Veil
• Vein
• Weigh/t
• Weird

Exception 2 (c-i-e):

• Ancient
• Conscience
• Efficient
• Policies
• Species
• Sufficient

Sidenote: this is not a random and arbitrarily bad heuristic, but one which is widely described as one of English's most valuable educational spelling rules. It's built into spelling softwares, national curricula, etc. English spelling is above-average wonky in terms of human-designed systems, but probably uncommonly regular in terms of emergent, evolved systems: this should give us some sense of how rough and error-prone many of our categorizations, taxonomies, and abstract nouns really are.

Now, can we try to better articulate or compress this set with a ruleset (that is, the extension with an intension)? Many have tried.

The most common variant comes from a late 19th C spelling book: i before e / Except after c / Or when sounded as “a” / As in neighbour and weigh. This is somewhat better—but what of “seize,” “weird, “height,” and many others?

A second variant is “i before e except after c when the sound is ‘ee’.” This, however, fails with “caffeine” and “leisure” (for exception type 1) and “policies” and “species” (with exception type 2).

We could try other possibilities, but our attempt would likely be futile:

In 1932 Leonard B. Wheat examined the rules and word lists found in various American elementary school spelling books. He calculated that, of the 3,876 words listed, 128 had ei or ie in the spelling; of these, 83 conformed to I-before-E, 6 to except-after-C, and 12 to sounded-like-A. He found 14 words with i-e in separate syllables, and 2 with e-i in separate syllables. This left 11 "irregular" words: 3 with cie (ancient, conscience, efficiency) and 8 with ei (either, foreign, foreigner, height, leisure, neither, seize, their). Wheat concluded, "If it were not for the fact that the jingle of the rule makes it easy to remember (although not necessarily easy to apply), the writer would recommend that the rule be reduced to 'I usually comes before e,' or that it be discarded entirely.”

This is only for words found in elementary school spelling books. 3,876 words. You know how large the average college-educated English speaker's lexicon is? 20-40,000 words, depending on your source.

Let’s add an incentive structure now, to introduce our surrogation mechanics. Recall that an incentive structure consists of several parts:

The bounty is the game's reward. The rules by which the reward is dispensed we'll consider its letter. This letter is an attempt by the designer to implement their intended spirit: the holistic, often vague goal the designer wishes to accomplish, and the holistic, often vague style of play the designer anticipates accomplishing this goal. Finally, there is the game's metric: the method for monitoring a player's behaviors, and determining an interpretation of reality which can be measured against the letter (that is, measured by the letter) to selectively dispense the reward. Such a system requires an evaluating agent or evaluating mechanism; in reality, its mechanism or agent will vary widely in its purview (programmed, legal, bureaucratic, or otherwise) to interpret the player's accomplishments against the game's spirit-conveyed-in-letter. (from “Surrogation Matrix” )

In diagram form:

Let’s imagine an institution with the goal of producing ad copy. To simplify the circumstances, we’ll suppose that their only goal is to produce copy where words in the purview of the “i before e” rule are spelled correctly, that is, receipt, policies, weight, and tier (e.g.) would be spelled as they are in modern English dictionaries. This is the institutional spirit.

This spirit will be measured by the “i before e, except after c” surrogate metric, or letter. The company does not want to have expensive human evaluators hand-checking every word with a dictionary, so it has implemented a very simple recognition algorithm to identify letter sequences according to the programmed surrogate metric (“i before e except after c”).

In this system, copywriters are rewarded (through pay bonuses) and promoted up the ranks based on their evaluation by this checking algorithm. Predictably, since the surrogate metric is the basis for reward—in other words, the real incentive structure, and not just the theoretical purpose, of the game—it will become the incentive structure of players. As those who spell “weird” w-e-i-r-d instead of w-i-e-r-d are repeatedly punished by the recognition algorithm, they will may be demoted to positions of less power and influence. They may choose to leave voluntarily, because they feel that, despite spelling their words “correctly,” and producing dictionary-perfect copy, they are losing prestige and recognition to those who produce dictionary-imperfect (but letter-perfect) copy. They may be asked to leave by the company, based on their poor performance reports. Or they may alter their behavior, intentionally spelling words “incorrectly.”

Here, the situation is so bared and obvious that it would be unlikely to play out in real organizations. However, there are certain qualities to this thought experiment not present in real life organizational oversight (“distant management”) which make unwitting surrogative reward of spirit-violators, and punishment of spirit-followers, more likely.

The first is that in our thought experiment there is an objective, demonstrable set of behavior (copywriter spelling) and set of spiritually intended behavior (dictionary spelling), such that performance results can be objectively appealed, and managers can quickly become aware of the possible discrepancies between their test and their desired game-function (spirit). In real management, this is not the case: there is neither an objective spiritual conformity criteria to appeal to—some ledger that contains a listing to "correct" behavior in every conceivable circumstance—nor is there an objective record of one’s own behavior. The only record—in ethnomethodology terms, the only account—of one’s behavior and performance left behind exists inside the scoring system. Observers may hold subjective intuitions about one performer playing “the right way” and another playing “the wrong way,” but the intuition holders’ ability to articulate and persuade superiors as to this spirit—and to defend the validity of their observations, judgments, and understanding of spirit—entail the limits of their ability to account in such a way. Practically speaking, this looks like dense he-said/she-said matrices that managers will struggle to maneuver, especially amidst inevitabilities like accountant bias, variable parsing competence, and strategic selfishness. There will always be value alignment problems between those who report or arbitrate spirits, and the larger organization’s goals, which is part of why routinized and “objective” surrogative metrics get introduced to begin with.

Further, we can imagine that, in our thought experiment, at least some of the members promoted up this ladder will be cognizant of the fact that they had submitted incorrect spellings to beat the checker, and might—now that they are hold a high position in the company, and have good incentive to maximize the company's profits—attempt to change the evaluation system, so that clients will complain less about all the irregular "wierd"s and "fiegn"s. System fixed! But again, the real world is more complicated.

For instance, in academia, many of the frameworks and approaches we on Pfeilstorch would consider "fake" (e.g. conceptual analysis, uncritical operationalization, survey-based studies) persevere. This is both because it is less obviously the "wrong" way—e.g. the failures of conceptual analysis can be unintuitive and difficult to grasp, and requires reading lengthy arguments about its problems—and because there is not much skin in the game for practitioners. Should the system at large keep playing ball, and awarding grants—and if there is no common knowledge, to usher in shame—there is no behavioral incentive to change. (In other words, there are no complaining clients, because first, there are no clients, and second, there are no dictionaries.) Because these academics, once tenured and supervising, will wish to continue their legacy, and not disrupt their own belief about the correct approach to philosophy, they will reward students who affirm their preheld biases. Students who grate against the status quo, and identify the pitfalls of conceptual analysis, will find themselves either in an uphill battle of persuading their mentors and superiors that their entire lifework is fundamentally flawed (unlikely) or will, frustrated and riddled with cognitive dissonance, drop out of the program. A similar situation is found in bureaucratic corruption: a corrupt police commissioner is more secure if he promotes similarly corrupt henchmen, since they are equally complicit (cannot rat him out) and are willing to enact his corrupt political will.

In other words, as knowledgeable insiders slowly leave the field (or choose never to join it in the first place), psychology will become increasingly dangerous and destructive until its public credibility collapses entirely. This process has been with the discipline from the beginning; academic psychologists Yoel Inbar and Michael Inzlicht report multiple occasions of "bright undergraduates" voicing complaints similar to Yarkoni's, and we can only imagine that psychology's inability to convincingly answer such concerns discourages those with the foresight to see it from entering. In other words, we have both a selection problem and a self-selection problem. We can call this "perverse selection"—the selection mechanism, of course, is perfectly standard, but the taste preferences are perfectly perverse, at least in light of the innocent spirit.