Suppose I am writing a poem, and have arrived at a first draft with which I am fairly pleased. As I read over it once again, a particular word catches my attention: it isn’t the worst word that could appear at that point in the poem, but nevertheless it seems like a placeholder, a word I put there temporarily in order to be able to move on and complete the rest of the line in which it appears. There must be a better word, one that will pull its weight instead of merely taking up space.

I write a few candidate words in the margin of the poem, and try writing out the line with each possible substitute in place. Eventually I decide that the entire line within which the word appears is weakened by it, and none of the putative substitute really resolves the deficiency. An alternative line suddenly comes to me; it seems immediately pleasing and right, and I make the substitution there and then without further deliberation.

Rather later on, after several cycles of alternating self-satisfaction and an insistently recurring mild nausea, I return to the poem, cross out the clever and charming substitute line, and replace it with the original. Later still, I change one of the words adjacent to the word that originally bothered me, and it then seems that the line is right enough, that it can be left alone.

If I am W. H. Auden, I then return to the poem a decade or so later and wantonly vandalize it, because my “late” aesthetic considers my “early” aesthetic a disastrous embarrassment and cannot bear to let any of its triumphs stand. The later Auden’s sense of what made a good poem good had changed to the point where he could no longer recognise what was good about his early poems, and so could not make them better. The alterations he made in this mood were attempts at damage limitation, the purpose of which was to dilute the venom of a rhetorical potency he considered anti-human. The result was mutilation: style cannot be subordinated to the demands of civility.

I do not think that there is any difference in kind between the question “how can you say that one poem is better than another?” and the question “how can you say that one version of the same poem is better than another?”. Auden’s inability, after a certain point, to make his early poems better was due to the fact that the investigation with which he had started, an investigation into what makes good poems and hence into what makes one poem better than another, had come to an end. He was investigating something different, namely the problem of how to fashion a civil discourse that would not promote or provoke atrocity; and the early poems’ wilful embrace of death, their passion for the real, was not commensurable with the aims and axioms of that investigation. I speak of axioms advisedly: the poem “The Shield of Achilles” explicitly opposes the “axioms” of civility (of “any world where promises were kept / or one could weep because another wept”) to those of atrocity (“that girls are raped, that two boys knife a third”; that the victims of barbarity will have “died as men before their bodies died”). The question of whether an investigation oriented by the axioms of civility could identify good poems is not in my view satisfactorily resolved by the later Auden.

The process of writing a poem is investigative: there is a “problem space”, which is that opened up by the poem’s initial gestures, and there are better and worse poems to be found within that space. It is not necessarily always a process of steady refinement: there are a variety of strategies for getting around. Departing from a known good in the direction of the very worst may sometimes be the only way of getting out of a local plateau in which there are at best only adequate poems to be found. But the criteria determining what is “better” and what is “worse” are themselves immanent to the problem space, or rather to the investigation which opens it up. In a sense, the answer to the question “what is a good poem” or “what makes a poem good” must always be a poem, or a group of poems through which the path of an investigation can be traced.

There is a parallel here with the problem of intelligence: how can we say that one intelligence is greater than another? Again, this question is not different in kind to the question of how I can judge my own intelligence to be functioning with greater or lesser intensity at any given moment. If it is possible for me to be stupid and muddle-headed in the morning, and full of wits and powers in the afternoon, then it is presumably possible for one person to be more or less stupid and muddle-headed, or full of wits and powers, than another, not only at some specific moment (some people are morning people, others are not) but over an arbitrarily lengthy stretch of time.

Now suppose that, as with the question of what makes for good poetry, we tie the question of how we can discern greater or lesser degrees of intelligence to a particular problem space, with its own immanent criteria of success or failure, that is itself opened up by an investigation, an aleatory and open-ended project. Once this is done, we can start to give an answer to the question of how we can say what constitutes intellectual greatness or superiority. The answer must always be given in the form of an observation of some intelligence in action, or some series of acts or occasions of intellection through which the path of an investigation can be traced.