I’ve received some helpful feedback on my recent post on Fifth Amendment protection for smartphones, and it has persuaded me that my Fifth Amendent analysis was partially wrong. I’m now not sure if the court was right in the SEC v. Huang case. But I think that if the result was right, it was right for the wrong reason. This follow-up post offers my latest thinking.

First, a recap. In the case, the government is seeking an order compelling defendants to disclose the passcode they used on company-owned smart phones. In my earlier post, I focused on how the foregone conclusion applies. The court concluded that the foregone conclusion doctrine only allowed the order if the government knew the existence and location of the incriminating records with reasonable certainty. Because that wasn’t known, the court reasoned, the doctrine didn’t apply. I argued that this was wrong, and that the foregone conclusion doctrine allowed the order so long as the government knew that the defendant had possessed that phone. That was known, so the doctrine applied.

That was yesterday’s post. But since then, Jonathan Mayer has persuaded me that the court was right that the foregone conclusion doctrine doesn’t apply — but for a different reason. In Huang, the SEC wants the defendants to respond to this directive: “Identify the Passcode for the smartphone that you used during the course of your employment.” The parties, the judge, and I had focused on whether and how the foregone conclusion doctrine applies.

Jonathan argues that this was the wrong question. The foregone conclusion doctrine can’t apply because the government isn’t asking the defendants to produce anything. The doctrine only applies to acts of production. But the government has sought an order here that the defendants have to say something, not turn something over. Because the foregone conclusion doctrine only applies to acts of production, not direct testimony, it can’t apply in the Huang case.

After thinking it over, I tend to think that’s right. The doctrine applies only to acts, not direct testimony. So I now agree with Jonathan on this point: When the government seeks disclosure or a password, the government is seeking a testimonial statement and the foregone conclusion doctrine isn’t relevant.

It’s worth considering why the foregone conclusion doctrine applies to the testimonial aspects of production but not to direct testimony. I would think the reason is that the government can know the testimonial part of an act of production but seek the act for non-testimonial reasons, while that can’t be the case for pure testimony. If the government knows you have a have tax records in your possession, and it wants to look at them to prove that you committed a crime, the government benefits from your turning over the records not because of the testimonial part (which it already knows) but because it then enables them to look through the records. The government doesn’t gain anything testimonial. On the other hand, when the government seeks testimony, it is necessarily gaining something testimonial — the subject of the testimony.

We’re not done, though. The statement sought is testimonial, but is it also incriminating? I’m still uncertain of the answer, although it may be only because I haven’t delved into all of the cases on incrimination. (I teach the Fifth Amendment, but I don’t live it in the way I do the Fourth. So I’m less confident in this area.) I’ll explain my uncertainty and invite readers to make their own call.

In Hoffman v. United States, 341 U.S. 479, 486-87 (1951), the Court stated that answers are incriminating if they “would in themselves support a conviction . . . [or] would furnish a link in the chain of evidence needed to prosecute” the defendant. At first blush, it’s easy to say the password is a link in the chain. If there is evidence on the phone, handing over the password is the way to getting to the evidence. It’s not evidence itself, but there’s a direct causal link to the evidence. (True, it’s possible that the passcode itself might be evidence — imagine a passcode like “insidertradingisfun.” But I’m assuming for now that the passcode is just a 4-digit number that isn’t itself of evidentiary value.)

If this analysis is right, then the password is incriminating because it provides a link to the evidence. The government could grant the defendants immunity, but it would need to be use and derivative use immunity — that is, immunity not just from the actual testimony but from what the testimony would reveal. See Counselman v. Hitchcock, 142 U.S. 547, 585 (1892). The defendants should win. That’s where Jonathan comes out, and it might be correct.

But I’m not sure. Here’s my question: Does the “link in the chain” test include a merely causal link — that is, a link in the chain to the evidence? Or does “link in the chain” mean that the testimony was part of the evidence of guilt but not enough to prove the entire offense — that is, a link within the body of evidence? If testimony is solely of value for its causal connection to evidence, and it has no evidentiary value itself, is the testimony incriminating?

As I understand things, the “link in the chain” test started in cases where the testimony was a link in the chain in the latter sense, as part of the evidence of guilt. The Court ruled that a statement doesn’t have to be a complete confession to be incriminating. Saying something that gives the government some evidence of guilt — a link in the chain — is enough. See, e.g., Blau v. United States, 340 U.S. 159 (1950) (in a prosecution for advocating the violent overthrow of the United States government, being forced to answer questions about the target’s understanding of and membership in the Communist Party was incriminating because it was a “link in the chain” of evidence supporting the conviction). As Hoffman put it, citing Blau: “The privilege afforded not only extends to answers that would in themselves support a conviction under a . . . criminal statute but likewise embraces those which would furnish a link in the chain of evidence needed to prosecute the claimant for a . . . crime.” Whether the “link in the chain” concept also extends to merely causal connections might have been addressed in Doe v. United States, 487 U.S. 201 (1988), but the Court punted on the issue. See fn. 5 (“Because we ultimately find no testimonial significance . . . , we need not, and do not, address the incrimination element of the privilege.”)

I’m puzzled about this question in part because of the discussion about incrimination in Fisher v. United States, 425 U.S. 391 (1976). In Fisher, the Court was considering the Fifth Amendment implications of issuing a subpoena to a taxpayer for the tax records that his accountant had created for him. That wouldn’t be testimonial, the Court suggested, because the testimonial parts of production would be a foregone conclusion. But then the Court added an alternative holding. Even if complying with the subpoena would be testimonial, it wouldn’t be incriminating:

[A]ssuming that these aspects of producing the accountant’s papers have some minimal testimonial significance, surely it is not illegal to seek accounting help in connection with one’s tax returns or for the accountant to prepare workpapers and deliver them to the taxpayer. At this juncture, we are quite unprepared to hold that either the fact of existence of the papers or of their possession by the taxpayer poses any realistic threat of incrimination to the taxpayer.

As for the possibility that responding to the subpoena would authenticate the workpapers, production would express nothing more than the tax payer’s belief that the papers are those described in the subpoena. The taxpayer would be no more competent to authenticate the accountant’s workpapers or reports by producing them than he would be to authenticate them if testifying orally. The taxpayer did not prepare the papers and could not vouch for their accuracy. The documents would not be admissible in evidence against the taxpayer without authenticating testimony. Without more, responding to the subpoena in the circumstances before us would not appear to represent a substantial threat of self-incrimination.

I interpret that passage in Fisher to suggest that the “link in the chain” test is about evidentiary significance, not causality. If the test were causality, then the discussion in Fisher wouldn’t make sense. Handing over records that might contain evidence of your crime is obviously incriminating in the causal sense. It enables the government to access the records, creating a causal link. But Fisher reaches the opposite conclusion by instead talking about the evidentiary significance of the testimonial act. It focuses on what testimony would be admissible in court (and thus evidence against the defendant), and what testimony on its own looks suspicious.

Apply that approach to compelling a passcode, I would think that the question is whether the defendant’s saying the password is itself evidence of guilt. This seems tricky to me. Following Fisher, it’s not illegal to know a passcode. Absent unusual circumstances, the password itself won’t be evidence — it’s just a path to evidence, like turning over the documents in Fisher. If the passcode were a text password, and could be something like “insidertradingisfun,” then that could be evidence. But I’m assuming that the passcode in Huang is just a short number, like 8256 or 2951. And if that’s right, then perhaps Fisher suggests that the password likely wouldn’t be incriminating. On the other hand, perhaps the analysis in Fisher is distinguishable because the defendant did create the password and is vouching for its authenticity? Or on the other other hand, perhaps the password is self-authenticating if it works? I’m not sure.

Where does that leave us, for the three readers still reading? In Huang, I agree that the government is seeking testimony, but I’m not yet sure if the testimony is incriminating. So I’m not sure which way the case should come out, as it hinges on the meaning of the “link in the chain” test. If I can figure that out, I’ll either add an update below or write a new post.

I said at the top that my previous post was partially wrong, which suggests that it was partially right. What’s the “right” part? My analysis was correct, I think, for the more common case when the government seeks a subpoena or order requiring the suspects to enter in their passcodes in a way that will unlock the phones. In that case, the government is seeking an act instead of testimony. The foregone conclusion doctrine should be satisfied and the order allowed when the government already knows that the person possessed the phone. So my earlier approach was correct for the case of “enter the code,” but the analysis above is my thinking at this point for the case of “tell us the code.”