Martin Geddes

Can I record that, please?

Etiquette. Beyond separating us from our less well-mannered primate cousins, etiquette is an essential development milestone within any technology adoption cycle.  Remember the early days of email when people would type IN ALL CAPS or developers thought the HTML <blink> tag was cool? (Seizures, anyone?) It took us a good while to gather our collective sea legs on the web and sort out good behaviour from bad.

As technology advances, new etiquettes must emerge. Without a common behavioural framework, adopting new tech can be intimidating and lead to awkward social moments, which the majority of us (self included) wish to avoid.    

As Hypervoice pioneers, we have had a central seat to watching new etiquette issues arise.  One that we’ve found profoundly interesting is the paradox of asking someone whether it is OK to record a conversation. What is the right etiquette, and how can our tools work to support good practises?

In an earlier article, we discussed the three core ethical values of Hypervoice technology: responsibility, respect and reciprocity. Any recording of call data, and subsequent use of that data, should comply with these principles.

Today, we see many violations of these principles, with the default being a highly asymmetrical power relationship - especially between enterprises and their customers. Every time you dial a call centre, and you hear “this call is being recorded for quality and training purposes,” imagine if you were then told “and you’re welcome to record this call too, a copy is being kept in your account for you to play at will.” That would be reciprocity! Knowing the relationship was reciprocal would also help us to avoid the kind of catastrophically disrespectful experience that Comcast famously inflicted on its customers. (5 million listeners and counting. Warning: it’s excruciatingly painful.)

So, when exactly is it okay to record? The paradox arises when the “contract” we have between participants is itself a verbal one and has happened before consent has been actually given. How do we record and non-repudiate that assent that was given?  Go back in time?

The conversation typically goes like this. People are joining in a conference call, and there is general chit-chat that you don’t want to record. Eventually you have a quorum of participants, and the meeting organiser asks if everyone assents to have the call recorded, say for your internal note-taking purposes and won’t be shared with anyone else. THEN you hit the “record” button, and off you go. The whole of the suggestion to have the call recorded, and for what purpose, and the assent to that, was all lost as it was prior to the call recording being turned on.

This prompts an interesting user experience design issue. Should we have “provisional call recording” to capture the request? Should the whole call be recorded by default from the start, but automatically discarded unless someone actively engages in a process to retain it? How can you locate any verbal assent to call recording in the Hypervoice call record as specific metadata? How should that be shared with participants after the call?

This hints at a much bigger issue around personal and corporate data retention policies and preferences, particularly with the global workforce we have today. In many ways, this verbal assent to call recording is the easy case. It’s not hard to conceive of extreme scenarios: you have person 1 who is a citizen of country A, employed by a company in country B, living in country C, but presently in country D, using a SIM card issued in country E, and is communicating with person 2 respectively associated with countries F, G, H, I and J, and they are using a service from country K, and talking about a person 3 who is a client based in country L. Whose call recording and data retention policies apply?

If widespread use of computer assistance in voice conversations is going to be adopted, then new technology will have to be created to automate most of these control processes. Conversations with close business associates will be recorded by default. Those with clients may be negotiated at the outset of the call as you enter the conversation space, and re-negotiated if needed. We may need as much machine learning technology to manage our privacy as we do for tasks like speech recognition and analysis.

Although the etiquette is yet to emerge, my personal view is that we should view this as a natural part of the adoption lifecycle, not a barrier. What we have found consistently is that the ability to have perfect recall of our conversations trumps the awkward moments. The benefit is so great that we are willing to stumble and figure out how exactly to phrase our request and, more importantly, when to make that request.