Earlier this month, my email and social media alerts starting going off when friends let me know a pseudonymous number-cruncher had "proved" Godwin's Law.

This struck me as odd, given that I designed Godwin's Law a quarter-century ago in a way that (I hoped) would make it insusceptible to scientific proof or disproof—at least by anyone who interpreted Godwin's Law as a prediction.

Although I'm proud of my career-spanning work on internet rights and freedoms, my biggest claim to internet notoriety is my little social experiment, Godwin's Law, which I crafted back in the days before there was large-scale public access to the internet.

Based on my own early experience of online arguments, I had come up with this mock "law," which was meant to have the sound and seeming inevitability of a law of physics or mathematics: "As an online discussion continues, the probability of a comparison to Hitler or to Nazis approaches 1."

I admit to being a bit of a prankster about this — I knew as a writer that if I could say something memorable about internet culture it was entirely possible for the memorable thing to take on a life of its own, propagated by the internet itself. After a bit of judicious promoting by me in the early internet discussion forums (notably Usenet), Godwin's Law took flight on its own in the early 1990s. Like a smartphone alarm I've forgotten to turn off, it pops up startlingly from time to time when I least expect it.

This happened on May 4 when it was announced that "CuriousGnu" – a blogger who shares with me an ongoing curiosity about numbers and statistical data –had blogged that "78% of Reddit Threads With 1000+ Comments Mention Nazis".

This finding didn't surprise me, exactly — when I came up with Godwin's Law, I based it to a large degree on my experiences in the 1980s with computer bulletin-board systems. Reddit, which has hundreds of millions of users, is in many ways like those 1980s bulletin-board systems — only, of course, millions of times larger. So one might guess than anything I'd seen on systems with dozens of users would certainly occur on systems with tens or hundreds of millions.

But CuriousGnu, who is righly cautious about overgeneralising from a few passes at Reddit's massive and admirably public dataset, was careful to state expressly that he was not attempting to prove or Godwin's Law, despite how his analysis is being reported.

Earlier analysts have not been so circumspect; a physicist named Travis Hoppe argued only last year that his analysis of Reddit data disproved the law. Like CuriousGnu, Hoppe likely surpasses me in mathematical skill, but (as I told him when he asked me about it on Twitter), the purpose of Godwin's Law was never to be predictive — instead, I designed the law to create a disincentive for frivolous or reflexive Hitler or Nazi comparisons so that, when we do feel compelled to make them in our arguments, we are more likely to be mindful about them.

I designed the law to create a disincentive for frivolous or reflexive Hitler or Nazi comparisons so that, when we do feel compelled to make them in our arguments, we are more likely to be mindful about them.

The internet has been shaping an increasingly international culture and collective memory — with the Holocaust, just as with other countless human atrocities, we have a moral obligation to "never forget". My view, which I've held for many decades now, is that glib and frivolous invocations of Hitler, or Nazis, or the Holocaust, are a kind of forgetting.

I'd like pretend that Godwin's Law is somehow applicable only to the internet, but of course followers of UK politics will have noticed that Hitler and Nazi comparisons have surfaced alarmingly this spring, most notably from two former mayors of London: Ken Livingstone and Boris Johnson.

Livingstone had stated publicly that Hitler at one point "was supporting Zionism" — a factually insupportable claim — and Johnson declared that Hitler's efforts to conquer Europe are reflected in the European Union, which he has called "an attempt to do this by different means."

What is one to make of these determined efforts by politicians of different parties to exemplify Godwin's Law? Personally, I can't be happy about them — I had hoped participants in public debates would grow less inclined to speak thoughtlessly about the Nazis and the Holocaust. In that sense, nothing would please me more to find that Godwin's Law, as any kind of predictive principle, could be "disproved" over time.

But the fact is, I designed Godwin's Law not to be predictive, but to be "memetic" — not to show that debates would invariably become overheated but to spur debaters to invoke history mindfully, with deeper analysis rather than with glib allusion, because that's the way for a speaker or writer to show that he or she is not taking the easy rhetorical path.

In order for the law to function this way, it needed simultaneously to "seem" scientific and yet function as a kind of negative inspiration (in effect, it hints at an ethical rule, not a scientific principle). And even if Godwin's Law does not always succeed in inspiring mindfulness, I hope it functions at least occasionally as a kind of unexpected "smart alarm" in today's heated political debates.

Mike Godwin is General Counsel and Director of Innovation Policy at R Street Institute.