TL;DR: Feet are arbitrary conventions in English verse, so you can scan the poem however you like, but Frost’s description of the metre is simpler than Gillespie’s.
The linked blog post (by poet Patrick Gillespie) describes two theories regarding scansion:
Most people divide lines into feet when they read.
In this division, most people prefer feet with rising stress (iambs and anapests) to feet with falling stress (trochees and dactyls).
I think that these theories are idiosyncratic to Gillespie, and that although we might find them useful to bear in mind when reading Gillespie’s own verse, the facts about English scansion are as follows:
Most people don’t divide lines into feet as they read: if they can do it at all, they have to do it as a separate, conscious, analytical step.
Division into feet is arbitrary and need not be based on rising stress: it can be based on falling stress, on the poet’s intention if known, or on parsimony of description.
How do most people scan poetry?
Here is how most of us will read the line:
Others | taunt me | with ha|ving knelt | at well-curbs
The implicit claim here is that “most of us” will insert these foot divisions into our reading. Now, Gillespie is himself a poet, and has written many detailed articles about scansion of poems (see for example this analysis of Donne’s Holy Sonnet IX), so I have no trouble believing that he divides lines into feet when he reads them. But I am quite sure that for most people, who are neither poets nor critics of poetry, this is not the case.
The reasons that I am confident about this are (i) I personally don’t divide lines into feet when reading poetry, I have to work out the scansion separately if I need it; (ii) “verse can be understood and appreciated by one who knows nothing of foot-structure or scansion” (C. M. Lewis; see below); (iii) scansion does not come naturally but has to be taught, with some difficulty, to children in school; (iv) a lot of English verse does not seem to have feet as such: in particular, children have no trouble with accentual verse like:
Oranges and lemons
Say the bells of St Clement’s
Is English an iambic language?
The English language is a naturally Iambic and Anapestic language (meaning that the language prefers a rising stress). In nominal phrases, we don’t normally say the car, we say the car. […] Also, all of our prepositional phrases prefer a rising stress (iambic or anapestic). […] This feature of the language has nothing to do with regional or dialectal inflections. It is simply the way our language works. It’s the reason Iambic meters, rather than Trochaic meters, are the dominant meters of English poetry.
There seems to be a missing step in the argument: just because we say ‘the car’, it does not follow that ‘car’ goes at the end of the foot rather than the beginning of the next foot. Gillespie must be applying some unstated principle. But what can he have in mind? The discussion of prepositional phrases suggests that it is something like “short grammatical units like ‘to the light’ should not be split by foot boundaries (if possible)”.
The easiest way to see that this argument is unlikely to be universally convincing is to quote a poet who applied a different principle and got opposite results:
Every foot has one principal stress or accent, and this or the syllable it falls on may be called the Stress of the foot and the other part, the one or two unaccented syllables, the Slack. […] for purposes of scanning it is a great convenience to follow the example of music and take the stress always first, as the accent or the chief account always comes first in a musical bar. If this is done there will be in common English verse only two possible feet—the so-called accentual Trochee and Dactyl, and correspondingly only two possible uniform rhythms, the so-called Trochaic and Dactylic. […] These are the facts and according to these the scanning of ordinary regularly-written English verse is very simple indeed and to bring in other principles is here unnecessary.
Gerard Manley Hopkins (1919). Poems, p. 1. London: Humphrey Milford.
Lacking any kind of universal principle that might decide between Gillespie’s “stress at the end of the foot, like grammar”, and Hopkins’ “stress at the beginning of the foot, like music”, we’re forced to conclude that the choice is arbitrary. Here’s the prosodist C. M. Lewis putting this conclusion plainly:
Feet are arbitrary and phantom concepts, and this description of the line does not touch the really vital fact about its rhythm,—the underlying regularity of its time-scheme. […] there is much dispute about the scansion of the line quoted. It seems to me, however, that all such disputes are teapot tempests. Since feet are merely matters of convenience, and rival systems of scansion merely different devices for attaining convenience at the expense of scientific accuracy, it seems useless to spin theories about them.
C. M. Lewis (1907). The Principles of English Verse, p. 43. New York: Henry Holt.
If feet are arbitrary, why use them at all?
A description of rhythm in terms of feet can be very concise. Compare “iambic pentameter” with:
a decasyllabic line on a disyllabic basis and in rising rhythm (i.e. with accents or stresses on the alternate even syllables)
Robert Bridges (1921). Milton’s Prosody, p. 1. Oxford: Clarendon.
There is a widely understood set of terminology about feet, which is convenient for writing about the metre of poems. Compare “the first foot is a spondee” with “the first and second syllables of the line are stressed.” C. M. Lewis again:
In the foregoing paragraphs I have avoided all mention of the word foot. It would often have been more convenient to use it, and I have been put to some circumlocution to escape from it; but I wished, even at the risk of awkwardness, to make it clear that verse can be understood and appreciated by one who knows nothing of foot-structure or scansion. Feet are not organic elements of rhythm. […]
The convenience, however, of description by feet is so great that it is idle to object to it on theoretical grounds. Even the fact that our names of feet are borrowed from the ancients, and that the rhythm of their verse was measured by the length of syllables instead of the intervals between accents, does not forbid us to use those names in description of our own verse; for they have become so fixed by long misuse in our vocabulary that the misuse has acquired a prescriptive right to recognition; and nobody has devised a better system.
Lewis (1907), p. 39.
Many poets composed their poems within a metric tradition that used feet, and so to understand their rhythmic construction it helps to determine how they intended their poems to be scanned within that tradition. When confronted with a line like:
All whom war, dearth, age, agues, tyrannies,
(from Donne’s Holy Sonnets VII) it’s helpful to know that Donne intended this to be iambic pentameter within a tradition that had rules about allowable substitutions, and that what he was likely to have had in mind was therefore:
All whom | war, dearth, | age, ag|ues, tyr|annies,
The parsimony principle
If we accept that the benefit of describing metre in terms of feet is primarily one of convenience, then we can compare candidate scansions to see which describes the rhythm most simply (in the information-theoretic sense).
In the case of ‘For Once, Then, Something’, Frost’s intended description of the metre would seem to be “trochaic tetrameter with dactylic second foot”, whereas Gillespie’s description is “tetrameter with cretic, anapestic, iambic, and amphibrachic feet respectively”. To remember Frost’s description, we need only remember two types of foot and the location of the exceptional foot, but to remember Gillespie’s requires four types, two of which are rare in English verse. Which is not to say that Gillespie’s description is wrong, since after all it accurately gives the positions of the stressed and unstressed syllables; it’s just that Frost’s description is simpler. Indeed, Gillespie acknowledges this when he points out that he needs an ultra-rare molossus in the variant line 12, whereas Frost only needs a common spondee.
Is trochaic verse clunky?
Some of it is, notably Longfellow’s ‘Song of Hiawatha’, but then is Longfellow’s iambic verse really any better? Here are four trochaic poems that seem to me to flow smoothly:
John Cunningham, ‘Fortune to Harlequin’:
From my favour, sense rejected,
Fools by Fortune are protected:
Fortune, Harlequin, hath found you,
Happiness will hence surround you.
William Shenstone, ‘The Princess Elizabeth’:
Malice never taught to treasure,
Censure never taught to bear;
Love is all the shepherd’s pleasure;
Love is all the damsel’s care.
Edgar Allen Poe, ‘The Raven’
Ah, distinctly I remember it was in the bleak December;
And each separate dying ember wrought its ghost upon the floor.
Eagerly I wished the morrow;—vainly I had sought to borrow
From my books surcease of sorrow—sorrow for the lost Lenore—
For the rare and radiant maiden whom the angels name Lenore—
Nameless here for evermore.
A. E. Housman, ‘Reveillé’:
Clay lies still, but blood’s a rover;
Breath’s a ware that will not keep.
Up, lad: when the journey’s over
There’ll be time enough to sleep.