In 1966 all of the US currency was withdrawn from Vietnam and our
government printed Military Payment Certificates.How did he know and how did Bulldog know? I guess he would
have just figured that if I went then Ted would go also. I had always been amazed how some of these guys know about everything that goes on. I guess it is good to not be surprised.As I walked over to Operations I saw my favorite C-54 crew chief.
— Once We Were Aces by Steve CampbellI heard that the alarm post near the sea had been quickly broken and
the tsunami warning couldn't be heard. I guess he would have
tried to pick up the glass broken by the earthquake. His body lay
face up on the floor in the middle of the kitchen, wearing work gloves
in his hands and a pair of trainers, the usual clothes that he used to
wear when he went for a walk. On the day I found his body, I
immediately went to the fire station and asked them to take his body
out from the house. However, they didn't come.
— The Great East Japan Earthquake The life given to me again by Ekuyo Abe"Who is that guy?" Aaron looked at the time. It was almost midday.
"A friend of Andrew's. All I know is that he trained with the SAS and
owes Andrew some kind of favour," Scott said. "I don't like to
question him too much.""I can understand that. I thought he was going to kill me."
Scott chuckled. "I guess he would have looked pretty scary. But you
can trust him, Aaron. Just make sure you jump when he says jump and
don't move till he gives the all-clear."
— Operation Alpha Papa by D. J. Stutley
The first two excerpts were written in the past tense. Why did the writers use the present tense form of "guess"? Because the reference time changed to the speech time as the writers was writing it? Seems to me it was not very likely. However, I think it is the case for the third excerpt as the guess is included in the direct speech.
If these uses of "guess" are conventional, then why use "would + perfect infinitive" constructions in the subordinate clauses? I can't find any counterfactual or hypothetical elements implicit in the contexts. I would just use "I guess he figured/tried/looked" instead. What is the contribution of the "would + perfect infinitive" constructions to the semantics of these examples (if it is not that of expressing counterfactuality)?
Best Answer
From a very strict point of view—say, the high-academic usage that obtained when I came of age in the middle of the last century—you are correct. The "formal" way of expressing this would be:
The "logic" behind this construction is that the WILL which is employed here is an epistemic: it expresses the speaker's inference. (As you are careful to remark, these clauses are complements of the verb guess = infer.) Here's that will in a present-RT context:
In the old way of doing things, when you make an inference concerning a past event you keep the modal in the present tense, because you are making the inference in the present, and you cast its complement in what I call the "sham perfect", employing the perfect construction as a past marker.
BUT—
There is a very-long-term drift in use of the modal verbs: the distinction between the present-tense and past-tense forms is gradually being redefined along lines of modality rather than lines of time-reference. This redefinition is very complex—the various modals are shifting at different rates, and the lines for each are being drawn in different places—but by and large what's happening is that past-form modals are assuming many of the uses traditionally assigned to their present-form versions, and the present-form versions are being restricted to narrower uses. Today we'd be just as likely to say:
Moreover, this evolution is occurring faster in colloquial English than in high-register formal English (as you might expect, since colloquial English does not have the same need as formal English for 'backward compatibility' with older uses). That's of particular relevance to your question, because all your examples emulate colloquial not formal use. What we see in your examples is a more advanced stage of modal evolution.
The process becomes clear if you compare the shoulds in your example to a modal which has already virtually completed this evolution—must. Historically, must is in fact a past-tense form which has entirely replaced its old present-tense form, mote, to such an extent that in all but the most formal registers must is felt to be incapable of expressing past reference—which is now expressed with the periphrastic had to. And if you replace your shoulds with musts (which in this context bear a virtually identical sense), there is no awkwardness at all: