I wrote a book arguing that we should be optimistic about our knowledge of the deep past: once we understand the methodological strategies historical scientists adopt, we should be confident that they will deliver the epistemic goods. Last month, Joyce Havstad wrote a piece in response (squee!) and in addition to saying nice things (double-squee!) she focused on two substantive points*: first, questioning just which sources of pessimism my arguments target (and whether I hit them); second, on the implied taxonomy of historical kinds underwriting my discussion of historical evidence. It’s a privilege to engage with people—especially someone as perceptive as Joyce—on these ideas. As she says,
“I’m really excited to hear what he thinks about what I think about his book”
So here’s what he (ur, I) think(s) about what Joyce thinks about his (my) book.
Let’s start with an odd caveat. Allow me to officially commit to the death of the philosopher. Infamously Roland Barthes argued that an author’s intent shouldn’t play a role in how we interpret their literary work. Similarly, I think that philosophers shouldn’t demand special authority over how their arguments are interpreted, or the ways in which their work is valued (or not). For one thing it’s a bit precious, for another it restricts philosophy’s potential use. Such attitudes tend to resist the thought that philosophical practice is primarily a social activity: something which I think is critical for producing good philosophy. One of course can disagree about what the best (or a good) interpretation of a piece of philosophizing is—whether you are the author or not—but once something is out there I reckon you should follow the advice of that song from Frozen. Hopefully I’ll manage to treat past Adrian in this vein.
Part of my strategy in Rock, Bone and Ruin is to identify three sources of pessimism about historical reconstruction and argue that they have been over-sold. Consider the thought that historical scientists cannot manufacture evidence as experimental scientists do. I draw an analogy between the kinds of things experiments can do on the one hand, and what historical scientists can do with modelling on the other hand (see chapters 9 and 10), suggesting that (to some extent) the virtues of the former are available to the latter. If historical scientists can manufacture evidence after all, then this rung of the pessimistic ladder is removed.
Joyce highlights a discrepancy between how the sources of pessimism are described in the book’s introduction and then how they are described further downstream. In particular, she highlights a switch between this from the introduction:
“Much information from the past has degraded or disappeared.”
And this quite different statement from chapter 5 which seems to replace it:
“We are unlikely to uncover further traces.”
Let’s call these source1 and source2. Source1 and source2 are surely different, so why do I apparently conflate them?
Perhaps I did something unsavoury to the proverbial dog on this one by letting the ghost of a previous version haunt the introduction. Let me explain what I mean. My partner Kirsten is a Newton scholar. She spends a fair bit of time examining differences between various editions and draft materials of Isaac Newton’s publications. Through that writing, we see him forever updating his work as statements are relabelled (from ‘principle’ to ‘hypothesis’, say), things are endlessly reworded, and so forth. Perhaps past Adrian was doing something like this, and due to an editing muck-up an earlier version lived on in the introduction. I’m just not as careful as Isaac. Maybe: having a quick look at the original first draft of the book from June 2015 (you heard me), the discrepancy Joyce notices is already present: if it is a ghost, it is an ancient one.
But perhaps there’s something more to be made of this discrepancy. Let’s attempt a rational reconstruction.
Source1 concerns how much information from the past is retained in the present; the thought being: not much. We can clarify source1 via a distinction of Elliott Sober’s. Sober distinguishes between information preserving and information destroying processes. In geology, subduction processes—when the edge of one tectonic plate slides under another— are information destroying as stratigraphy is distorted and the differences between layers is erased. In the worst cases, the original ordering of strata is unrecoverable. Fossilization is one example of an (often quite crappy) information preserving process. I think a reasonable reading of source1 is to say that historical processes, or at least those which matter to historical science, are more often information destroying than information preserving.
On this reading of source1, its connection with source2 is I think clearer. If most historical processes are information destroying, it is a short inferential walk to the claim that there won’t be many traces available for historical scientists to work with. Hence, we will be unlikely to find new traces: source2. With that clarification, let’s try reformulating the claim:
Because of the ubiquity of information destroying processes, we are unlikely to uncover further traces.
On this interpretation source1 is perhaps sufficient (but not necessary) for source2: it could be that we lack the requisite technology, understanding or will to uncover further traces, even though there is plenty of retained information. Regardless, we can see that despite their differences, source1 and source2 fulfil similar roles in the book’s argument: one underwrites the other. This differs slightly from Joyce's interpretation that in fact there are four sources of pessimism in the book.
Joyce worries that I haven’t actually provided reason to think source1 doesn’t hold, and further that this matters: specifically, without undermining source1, we end up with an agnostic position vis-à-vis historical reconstruction. As she says,
This uncertainty [concerning whether historical processes are information destroying or preserving] is what supports the agnostic position: there is at least enough opacity when it comes to the past to obscure the nature of our access to it, yet we are successful enough in recovering it, sometimes at least, for us to also be unsure about whether and how often we’re really missing anything significant.
Going with my construal of past Adrian’s claim—that history involves many history-destroying-processes—is Joyce right that I haven’t provided reasons for denial? Well, perhaps not directly. One of my aims in the book is to push localism and context-sensitivity about epistemology. I doubt generally speaking history is dominated by information preservation or destruction. However, I think there are three closely related arguments in the book that can be bought to bear in supporting optimism in this regard.
First, I think the scaffolded nature of historical reconstruction leads us to overestimate information destruction. It is only once various hypotheses have been explored that we can actually identify new evidence (see my discussion of the role of art in paleontology). Prior to reaching an evidential scaffold we cannot tell what new evidence shall arise. And so, I suspect, our judgements about information retention will be biased towards pessimism. That’s a pretty weak response (biases can be corrected – or over-corrected!), but we can consider the point alongside my second and third arguments, which point to epistemic developments: new understanding, as well as new technologies, techniques and tools, which reveal history to have been less information destroying than we expected (Ben Jeffares makes a similar point in response to Derek Turner’s pessimism).
I’ll put the second point briefly: our capacity to extract information from faded, decayed past remnants is sensitive to our background knowledge about those processes. And this knowledge is continually developing: if we sufficiently understand an apparently information destroying process we might discover it was information preserving after all. The history of paleontology is marked by such discoveries.
Third, information destruction is typically cashed out in tracecentric ways: information is imagined as being ‘contained’ in the remnants of past events. I think this is misleading (as we’ll see below, I *think* I can say this without being committed to anything too weird about the nature of information). To see why, let’s turn to Joyce’s second discussion.
Joyce picks up on a distinction I draw (in chapter 7) between two kinds of evidence. ‘Trace’ evidence involves drawing historical links between some contemporary remain (the trace) and a past target. Fossils are evidence of extinct critters because extinct critters are causally upstream of fossils, and we have good theories which explain how fossils form. ‘Analogous’ evidence is not causally linked in this fashion, but rather involves object and events which are produced by similar processes. Igneous rocks are not grouped together because they are ancestrally related, but because they formed when molten lava cooled. Joyce speculates that this misses another category: things that are not simply similar due to continuity in the processes which form them, but are the same.
I’m not entirely sure what to make of ‘same/similar’ talk (at least this version of past Adrian doesn’t think much hangs on it). I do think that biology contains a bunch of categories that combine analogous and trace categories (most obviously parallelisms), but I don’t think that is what Joyce is getting at. I think in essence the suggestion is that we should take analogous evidence—and analogous categories—seriously.
In some recent (currently unpublished) work on historical kinds, Laura Franklin-Hall distinguishes between two ways of categorizing with history. First, there are token-historical-kinds (which I think roughly aligns with categorizing things as ‘historical individuals’). This is a bit like trace evidence: we draw a common causal history linking the objects together. This blog might be like this: Extinct is a (fairly casual) institution whose identity is maintained by overlapping chains of continuity (topic matter, authors, website, publishing schedule, etc…). This post is a part of Extinct in virtue of being relevantly connected to those chains. Second, there are type-historical kinds. Here, we group objects together in virtue of their having undergone common (but separate) processes. Again, consider igneous rock: two pieces of igneous rock don’t count as such because they came from the same volcano, but because they were both formed by the relevant volcanic process. This roughly aligns with what I called ‘analogous’ evidence. Especially in biology, historical token-kindhood is taken much more seriously than historical type-kindhood (Joyce cites Paul Griffiths’ claim that homology plays a central role in biological categorization, which I think is the most convincing version of this position).
I take Joyce to suggest that we take type-kindhood much more seriously, and that we can do this by distinguishing between cases where similarity is coincidental and when that similarity is explicable by some combination of trace (common ancestry) and analogy (common but separate history). When similarities are explained by common but separate histories, it seems like there are reasonable grounds to say these are the same, not simply similar.
I’m not sure if I’ve quite got Joyce’s suggestion right, but with the analogy/trace (historical token/type) distinction under our belts we can go back to Joyce’s original criticism. Recall that her worry was that I haven’t provided reason to think historical processes aren’t often information destroying. I think notions of information destruction (and retention) is far too focused on traces. When we imagine information destroying processes, we typically think of it in terms of particular causal histories: fossilized bone scatters, losing original shapes and positions, and thus limiting our capacity to piece everything together. However, our reconstructive capacities are dependent on a host of things beyond just the available traces: analogy matters too. For instance, if we have many examples of that process-type—if we’re in a good position to understand the dynamics of fossil formation, disposition, and so forth—then much more information about the past might have been retained than we originally guessed. And this doesn’t end with analogy: there is also the coherency of our picture of the past, how it fits with other streams of evidence, and so forth. Even if historical processes work hard to hide their tracks, our powers to recover those tracks are I think often stronger than we realize.
This, I reckon, takes us some way towards an optimistic view.
I’ll close with two deep ambiguities in the book which Joyce’s criticism reveals. One concerns the relationship between traces and past information, another historical evidence itself. When I say that evidential sources such as analogy can mitigate information destroying processes, does this amount to saying that we can get more information from a trace than that trace itself contains? That seems like a pretty odd claim perhaps, although maybe if you’re willing to go somewhat constructivist about information then it might not be so weird. Regardless, given that we’re in the business of understanding historical science as it is in fact practiced (and indeed, my conception of ‘traces’ doesn’t turn on how we think about information—see chapter 3!), I’m not sure this is a difference that makes a difference.
The second ambiguity is perhaps more pressing. In arguing for optimism, does Adrian claim that historical scientists are successful despite their terrible evidential situation, or should he claim that we are wrong that historical scientists are in a terrible evidential situation. On the first reading, historical scientists in fact face an impoverished evidential situation, and make the best of it. On the second reading, we have misunderstood how historical evidence works, and their situation is actually not impoverished at all. I think the latter better characterizes Adrian’s view, but he doesn’t always do a great job of distinguishing them and (death of the philosopher!) indeed reading me (him) in the former light is, I think, legit.
It is such an honour to have kind, critical and creative folks like Joyce reading and interacting with my work. Responding to Joyce’s thoughts have led me to reconsider some of my argument and see them in a new light. As I noted above, philosophy is a social activity and gosh it’s a wonderful experience when such back-and-forths click productively, as I hope it has here…
*Joyce's critique also involves a fascinating problematization of my appeal to the metaphor of ripples in a lake, I haven't time to ruminate on this here, but I think it is nonetheless pressing.