FCC rules have the force of law. That means people who violate them are subject to a fine.

But suppose a company sells a product having a feature the rules do not plainly prohibit. Even a careful reader might not take the rules as alluding to this feature at all. The company might plausibly think its actions are perfectly okay. And, just to make it interesting, let’s suppose another part of the FCC had approved the same feature as being lawful.

Now the FCC’s enforcement arm claims the feature is a violation, and imposes a fine.

The General Mobile Radio Service (GMRS) is a category of two-way radios, usually more powerful and of better quality than the little FRS “family radios” sold everywhere. Users must hold individual licenses from the FCC. The radios must be certified by the FCC for compliance with applicable rules. One of those rules prohibits GMRS users from sending “coded messages.”

Several brands of GMRS radios come with a “voice privacy” feature that converts the speaker’s voice to gobbledygook. Any radio from any manufacturer with the same feature converts the voice back to intelligibility.

The FCC recently singled out one of these manufacturers for a $21,000 fine, claiming the radio was capable of sending what it called “coded messages.” Could the manufacturer reasonably have known in advance that the feature was unlawful?

The algorithm for the privacy feature, called “voice inversion,” is trivially simple: a circuit in the radio reverses the speech spectrum, exchanging high and low frequencies. Nothing else. The signal is not encrypted. No password or key is needed. The algorithm is well enough known to have its own Wikipedia article. The FCC notice that advised the manufacturer of the fine – a published document, available to anyone – has enough technical information for a skilled hobbyist to build a de-inverter. Those less skilled can buy a pre-built de-inverter on a chip. Or download free software that does the same thing.

The manufacturer might reasonably have concluded that a signal so easily recovered is not “coded,” as most people understand that term. Inverted, yes.  Possibly even “scrambled,” though in a primitive way. But to us – and possibly to the manufacturer – the word “coded” suggests a scheme by which parties expect to keep their messages absolutely indecipherable except to those to whom the key – the code – has been entrusted. Spy-novel stuff. Anyone trying to use voice inversion for real privacy would be in for an unhappy surprise.

Oddly, even the FCC – at least, the part of the FCC that certifies GMRS radios – seems to have accepted voice inversion as permissible. When the manufacturer applied for certification, it handed over technical diagrams that fully disclosed the voice privacy feature. The instruction manual, submitted with the application, explained how to use the feature. The FCC routinely approved the certification. That amounts to a finding of compliance, say the FCC’s own rules, based on the information in the application.

But the people in the enforcement wing of the FCC – the people who levied the fine – disagree.   They claim a feature can be in violation of the rules even if the radio is certified, and even if the FCC knew about the feature in approving the certification. To support this view, they cite a rule that empowers the FCC to revoke a certification, if it determines a product does not comply. But immediately following that is another rule on the procedures for revoking a certification. Those require, among other things, advance notice to the certification holder and a chance to respond. Neither one happened here. That left the certification unrevoked, and presumably valid, at the time of the supposed violations. Until now, that would have sufficed to establish compliance.

To be sure, as technology evolves, the FCC sometimes must change how it interprets its rules. There is a procedure for that, too: the FCC issues a public notice that alerts manufactures and others to the change. There was no public notice here. But the enforcement office nonetheless claims the public had fair warning about voice inversion – not once, but twice.

One instance, it says, was a 2007 Q&A entry in an FCC “knowledge database.” The entry indeed states clearly, if awkwardly, that “voice scrambling is coded messages” and prohibited by the GMRS rules. But the FCC made no announcement when it posted the entry. To find it, a person would first have to suspect its existence, and then go digging in the database with the right search terms.

In any event, for the FCC to assume public knowledge of the entry is flatly illegal. Its own rules bar the FCC from relying on any document that does not appear in one of four named publications. The knowledge database is not one of the four. That makes the Q&A entry inadequate notice as a legal matter, besides being a poor way to get the word out.

The other supposed public warning was a 2004 waiver order allowing another manufacturer to transmit GPS location data over a GMRS radio.   The GMRS rules restrict communications to “two-way voice messages,” plus tones for specified purposes. All data, including GPS location data, are prohibited. Accordingly, the manufacturer needed a waiver.

How, then, did a GPS data waiver warn the public against voice inversion, which does not entail data at all? True, the manufacturer had put the “coded messages” rule on the list it wanted waived. We don’t know why; perhaps it thought the GPS data format “encoded” the location coordinates. The FCC duly put that rule in the list that it waived, but it did not say why, either. The term “coded messages” does not even appear in the waiver order, except in the list of rules.

Now, though, the FCC has gone back to revise what appears to have been its original rationale. GPS-as-data is no longer central. The new thinking goes like this: GPS signals are “coded messages” because they require “necessary equipment” for reception – i.e., to “decipher” the location information. Voice inversion also requires necessary equipment for reception – the de-inverting circuit. Therefore, voice inversion is also a coded message, goes the reasoning, and hence against the rules.

This might have been a decent argument if the GPS waiver order had actually referred to “necessary equipment” as an element of the waiver, or even suggested that GPS signals are coded messages. But to our eye, the GPS waiver order is about data transmission and nothing else. It would have taken a high level of prescience to have studied this order back in 2004 and concluded that, in 2009, it would be read to ban a non-data feature such as voice inversion.

Compounding the confusion, the FCC okayed certification for some of the radios it now claims to be unlawful after the appearance of both the knowledge database item and the GPS waiver. If these had banned voice inversion so clearly as to justify a $21,000 fine, we can only wonder why that message did not reach the FCC people who handled the certifications.

The case concerns us, not just for the apparent injustice to one company, but because it undermines all of our efforts at FCC compliance. Most people regulated by the FCC work hard to comply. We lawyers do what we can to help, in part by keeping track of the statutes, the rules, and the published orders that explain and interpret and apply those rules. But we don’t read minds. Perhaps this is a one-time slip-up the FCC will soon correct. But if the FCC keeps re-interpreting the rules retroactively to mean things they don’t actually say, then even the best efforts at compliance cannot be enough. Regulatees had better keep a checkbook handy. They are going to need it.