Many have recalled the 2016 referendum; some have suggested a few minds have now changed; but (as far as I know) none have critiqued its basis: trying to resolve a multi-option problem in a binary vote on just one of those options  -  a classic case of artificial incompetence, the other AI.

In June 2016, there were four options ‘on the table’ - in the EU, EEA, CU or WTO - my press release of February 2016 spoke of multi-option referendums, like New Zealand's five-option vote in 1992.  But Brexit was binary, and a (small) majority didn’t like ‘remain’.  Maybe even bigger majorities didn’t like the other options.  As in Theresa May’s indicative vote on eight options, maybe there was a majority against everything.  And maybe, losing by just 52%, ‘in the EU’ was the least unpopular, and therefore the most popular.  Maybe ‘remain’ was the winner.

There were lots of things wrong with that referendum, not least the lies used during the campaigning.  But another huge error was to think that reducing four options ‘on the table’ to just one option on the ballot could be fair.