A post motivated by reading Dogmatism and Evolution by G and T Laguna:

Berkeley's subjective idealism has, amongst perhaps many others, a problem in accounting for regularities in experience. In Principles of Human Knowledge, we have the following (section 30):

The ideas of Sense are more strong, lively, and DISTINCT than those of the imagination; they have likewise a steadiness, order, and coherence, and are not excited at random, as those which are the effects of human wills often are, but in a regular train or series, the admirable connexion whereof sufficiently testifies the wisdom and benevolence of its Author. Now THE SET RULES OR ESTABLISHED METHODS WHEREIN THE MIND WE DEPEND ON EXCITES IN US THE IDEAS OF SENSE, ARE CALLED THE LAWS OF NATURE; and these we learn by experience, which teaches us that such and such ideas are attended with such and such other ideas, in the ordinary course of things.

The problem hidden here is that the phrase "in the ordinary course of things" needs cashing out, because it is typically not the case that as individual subjects, our experiences conform to the general regularities of nature (e.g we do not always ourselves hear thunder after seeing lightning - other conditions have to be in place). Mill pointed this out (ch 11 Examination of Sir William Hamilton's philosophy).

Now, of what nature is this fixed order among our sensations? It is a constancy of antecedence and sequence. But the constant antecedence and sequence do not generally exist between one actual sensation and another. Very few such sequences are presented to us by experience. In almost all the constant sequences which occur in Nature, the antecedence and consequence do not obtain between sensations, but between the groups we have been speaking about, of which a very small portion is actual sensation, the greater part being permanent possibilities of sensation, evidenced to us by a small and variable number of sensations actually present.

So what are these permanent possibilities of sensation, and specifically, what grounds them? Usually a possibility is something we think of as grounded in an actuality of some kind or another, and of course if that actuality ultimately has to be an independently existing object or fact, subjective idealism is done for. One solution is to propose that the actualities on which these permanent possibilities are grounded are actual sensations/experiences, which would save subjective idealism but at the following cost: possibilities become mind-dependent, and thus (at least if the mind is not God's) they lose their required permanence, since modality is thereby bounded by our capacities for imagining possibilities (which could and has changed, and also runs counter to the intuition that there are possibilities we as human beings have yet to consider).

So, as I see it, the issue for subjective idealism - and perhaps it affects all idealisms, although on that point I am not certain - boils down to the problem that it requires not only idealism to be true of the actual world, but also for all possible worlds. Is idealism for possible worlds coherent? After all, isn't it at least possible that idealism is false?