Why This Blog Is In Progress

Winter Break is over and it’s time to fire up the SCOTUS doctrinal mapping machine! I’ll start 2015 by looking back at my last post. No, this is not early-onset nostalgia for 2014. I first need to correct an error in that post.  My interest, however, extends beyond setting the record straight. Understanding what went wrong last time helps explain the very nature of this project and why I call this blog “In Progress.”

Let’s start with the post in question. If you don’t feel like re-reading, here’s the basic gist. I posited that Confrontation Clause doctrine post-Crawford could be visualized as a tussle between three camps: the bashers, builders, and swing vote. I argued that Justice Thomas was the swing vote “since he has voted with the majority in every single case in this line.” This map represents the argument (click for full size):

After putting this out into the blogosphere, I was lucky enough to hear back from two of the leading Confrontation Clause scholars in the country — Richard Friedman and George Fisher. Both professors kindly pointed out that Davis v. Washington opinion actually decided two cases (Davis plus Hammon v. Indiana) and that Thomas was the lone dissenter in Hammon. So it is decidedly not true that Thomas has voted with the majority in every post-Crawford case.

How did this error occur? The short answer is that I improperly assumed that Davis and Hammon could be plotted as a single data point.  No doubt this was a rookie mistake. Had I closely read the Davis opinions, I would have realized that Thomas joined the Court in denying Adrian Davis relief but dissented from granting Hershel Hammon relief. Yet I did not create the map by reading. Instead, I automatically generated a citation network using CourtListener and then coded that network using Supreme Court Database (Spaeth) data. And both CourtListener and Spaeth incorrectly identify Davis as a single datapoint. 

In fairness, CourtListener’s “error” is really not an error at all. The fact is that Davis/Hammon share the exact same citation so any citation-based network has to treat the cases as one. Put another way, no unique cite exists for the Hammon case decided by the Supreme Court on June 19, 2006. The practice of deciding two cases under a single caption will similarly confound every search engine — including Lexis, Westlaw, Fastcase, Ravel and so on.

On the other hand, Spaeth might have done better. That database is populated by human readers who theoretically code every case for many variables including decision direction (liberal vs conservative) and majority vote count. Under such a schema, the two cases really should be treated distinctly since the decision direction differed (the criminal defendant lost in Davis but won in Hammon) as did the majority vote-count (9-0 in Davis verus 8-1 in Hammon). In any event, it seems that the Court’s single-citation-for-two-cases practice should be added to the list of “watch outs” for scholars using Spaeth.

The Davis/Hammon hiccup stands as a good example of the kinds of challenges involved in applying network theory to understanding Court doctrine. Automatically generating citation networks is a powerful tool, but the Court’s complex hermeneutic practices require constant refining of the maps created by the tool. Now I call this blog “In Progress” because I am interested in publicly exploring this continual process of refinement. I thus happily invite others — experts or interested amateurs — to help me figure out when and why the tool gets it wrong or misses something.  Progress is a community effort.

With respect to the specific Davis/Hammon problem, I have an imperfect fix. This is what it looks like:

As a reminder, upward-facing triangles on the map represent opinions advocating in favor of the confrontation right/criminal defendant while downward facing triangles represent the opposite. Note how Hammon now appears to the left of Davis on the map and how Thomas’ opinion is marked as a 1-vote dissent rather than concurrence. Note also how I kludged Hammon‘s date to separate it from Davis — failing to do this would result in the datapoints appearing on top of each other.  

So the fix ain’t pretty, but at least it’s progress. Or so I hope. Comments more than welcome!

** Postscript ** Based on a comment from the SCOTUS Mapper’s very own Darren Kumasawa (see below), I must note that Spaeth actually DOES properly record Hammon and Justice Thomas’ dissent in it.  It just does not do this in the dataset we use (which is the citation-centered set rather than the issue-centered set). The existence of two datasets with not completely consistent data means that you still need to “watch out while using Spaeth”. However, I was quite wrong to imply that the good folks over at the Supreme Court Database failed to get their coding right on Hammon. My apologies! Once more, this is why I call this blog “In Progress”!!