One characterization of what happens with specialization in philosophy is that the strategy becomes that of collecting patches for an idea, and that certainly characterizes Plantinga’s theory of warrant. In my view, the sequence of patches reveals something important (i.e., defective) about the theory.
In brief, Plantinga begins by identifying warrant with whatever closes the gap between true belief and knowledge. This approach allows Plantinga to use examples that drive development of the theory in terms of whether those examples are examples of knowledge. All one needs to do is to say this: suppose the belief in question were true–would the person in question have knowledge?
When reading Plantinga’s theory from the point of view of thinking of warrant as whatever turns true belief into knowledge, I find myself allowing some of these steps where I wouldn’t allow them if I were thinking more in terms of the kinds of normative evaluations we intuitively make about beliefs: intuitive normative notions including warranted belief but also responsible belief, justified belief, reasonable belief, etc. But given the methodology, it would be a mistake to approach Plantinga’s theory in this way. Instead, I find myself being at least somewhat sympathetic with the following steps: identify warrant with proper function, require a design plan for things that function properly, and then build in both truth-conduciveness into the design plan and a friendly environment as conditions for warrant.
But here come the patches.
The first patch is to weaken the statement of what warrant is. Initially it was that which closes the gap between true belief and knowledge, but then it because that quantity which, when you have enough of it, and add true belief to it, you have knowledge. The first patch acknowledges something about the intuitive notions of warrant, rationality, justification, and responsibility–namely, that they come in degrees. If warrant is simply that which when added to true belief gives you knowledge, then the degree point seems to get lost. So a first patch is needed, and used.
It is not clear that using this patch puts us in the same position about the cases Plantinga uses to develop the theory. Once we start honoring aspects of the ordinary normative notions, it’s no longer clear that we can use the same cases Plantinga uses to draw the same conclusions. Instead of the simple procedure of asking whether by adding true belief to the story, knowledge results, we have to do something more complicated and harder to assess. We have to ask whether, assuming the belief is true, we’ve travelled down the road from true belief to knowledge in a given case. Now my reaction depends on my mood. In my charitable mood, I’ll give a lot. In my normal(!) mood, I’ll balk every time I can find something good about the belief from a purely cognitive point of view (besides the good involved in having a true belief). The result is that in normal moods, even the first step of identifying warrant with proper function is suspect.
Things get worse, however. A second significant patch is needed for Gettier cases, as Plantinga admits in 1996 in response to papers by Klein and Feldman in the volume of essays I edited. The patch involves distinguishing between a maxi-environment and a mini-environment, and Plantinga to this date still doesn’t have a good theory of a mini-environment (the latest attempt is a really a kind of defeasibility theory, but, I think, not adequately sensitive to the need to distinguish misleading from non-misleading defeaters).
This patch makes the sequence of steps in developing the theory intolerable, I think. Recall that we began thinking of warrant as an amalgam of whatever conditions are needed to turn true belief into knowledge. So, having been schooled in the Gettier literature, I understand Plantinga to want warrant to be a combination of the third condition, usually clarified in terms of justification, and the fourth condition, for which, I say tendentiously, the defeasibility theory is the only interesting game in town.
But now the first patch conflicts with this second patch. I can either think of the second patch as an additional condition for knowledge, or as an additional condition for warrant. If the former, then all my charitable inclinations noted above were inappropriate: the theory is that knowledge is warranted true belief in a suitable mini-environment. I should evaluate the examples used to drive the theory of warrant the way I’d evaluate any examples used to drive any normative theory, and then the very first move, from warrant to proper function, is suspect (lots of normatively positive beliefs involve improperly functioning equipment, just as good things sometimes, even if rarely, happen when machines malfunction).
That leaves taking the second patch as revealing an additional condition for warrant. But if I take it this way, it looks like ‘warrant’ is, once again, nothing more than a placeholder for whatever plugs the gap between true belief and knowledge, and when I think of it in this way, I no longer have any reason to think that warrant comes in degrees (the motivation for the first patch). Some conditions for knowledge may come in degrees, and if Hetherington is right, even knowledge itself does. But there’s no inference from either of these points to any point about warrant, conceived of as a placeholder in the way described. So either way, the patches don’t fit together.