I had an interesting discussion with my veterinarian yesterday after she had a look at our three dogs (they’re fine), covering the Hippocratic injunction to do good and avoid doing harm in the process of treatment.
It’s a fine idea, and I wonder if data practioners couldn’t learn something from that. There is considerable pressure for analysts to do something; to run data over algorithmic hill and dale, and arrive at an outcome or conclusion. That process isn’t necessarily easy, but it is familiar – we have the tools and know the process, and we can almost always generate some kind of activity and outcome.
But not infrequently, our first duty may be to assess whether there is something worthwhile to be done. If we suspect conditions are not right for success – that questions need to be refined, that unknown data context is present, that significant error or bias is in play, that’s a good time to ask our clients to re-assess their next steps.
I don’t know of any analytics oath such as doctors have – a kind of ethical backstop when patients demand more treatment than is sensible. And as analysts’ work is not considered immediately life-impacting, the situations are possibly different. But with analysis work increasingly engaged in medicine and other human activities, the “life impacting” line is no longer entirely bright and clean. And of course there is the question of investing our client’s time and money optimally. Perhaps an “oath” – a public declaration – to do no harm wouldn’t be such a bad idea. I know good analysts who quietly hew to that injunction already, but their ethical dedication doesn’t get the notice that it otherwise might.