It’s easy to see why tactical voting appeals to the politically minded. It’s a sort of clever hack to try to get round the twin problems of Britain’s first-past-the-post voting system and a split vote, usually on the left.

The formula is straightforward: in constituencies where it looks like the Conservatives might win with less than 50% of the vote, ignore your first preference and vote for the party that has the best chance of beating them. The trouble is that like so many life hacks, it always turns out to be much more difficult to get right than it might have seemed.

For example, getvoting.org, the website run by Best for Britain (a remain-focused campaign group) changed 83 of its constituency recommendations this week. Effectively, it was admitting that in the three weeks since its launch, one in eight of its instructions to voters were wrong. If these recommendations had been carried into a general election, of course, they would have had the opposite effect to the one intended – the “tactical” equivalent of sending your cavalry to charge the machine guns.

This ought to embarrass the people involved. When I was a city analyst, I was mortified on the occasions when I had to change even a single recommendation. But the website’s credibility is not so much the issue here. The problem is with the whole concept of tactical voting.

When the tactical voting sites were first launched in this election cycle, the most common criticism of them was that they were biased towards the Liberal Democrats, because they seemed to be recommending lots of votes for the party in constituencies where Labour had come second to the Tories in 2017. I think we can now see that this wasn’t the case – of the 83 changes made by getvoting.org, 81 of them are to switch the recommendation from Lib Dem to Labour, and one of them is a change of Kensington to “Too close to call”.

The problem was that the underlying model was based on a set of constituency-level forecasts, and those forecasts put the Lib Dems in second place in many constituencies because they were based on polling data from a period when the Lib Dems were polling very well. Over the first three weeks of the campaign, the polls have moved substantially, with Labour gaining support and the Lib Dems falling behind, and the getvoting.org model update (based on polling over the last week) reflects that. When other tactical voting sites get updated, they are going to have the same problem.

All well and good, but the fact that polls move during a campaign was hardly a surprise, was it? And with a fortnight to go, it’s by no means certain that the current recommendations are right, or that the sites won’t have to make another 80 changes before polling day.

Any one of those changes risks sending tactical votes the wrong way, because voters won’t necessarily return to check that the advice they got is still valid. The damage is likely to be small – the sad fact of first-past-the-post is that many of the recommendations are all but irrelevant because most seats are safe. But the risk involved is totally needless. And much more importantly, it has started an equally needless fight with exactly the people the tactical websites needed to convince – passionate Labour supporters.

The main consequence of launching based on October data has been to create a lot of ill-feeling for the entire concept of tactical voting. It would have been far better to spend the first month or so publicising the general concept of tactical voting, then coming out a couple of weeks before the election with a fairly reliable set of recommendations that were more likely to be taken in the spirit they were made.

The underlying problem seems to be that in the online political era, clever and enthusiastic people seem to choose projects based on what might go viral rather than what really needs to be done. Because nobody really needs one of these websites, let alone three or four competing ones. Anyone who can understand the concept of tactical voting and why they might want to do it is equal to the very easy task of doing their own research (the tactical.vote website even tells you how, in 200 words). People who don’t want to vote tactically usually have their own, often strongly felt, reasons for not switching to Labour or the Lib Dems.

The idea that there is someone out there who would vote tactically if they could just get a convenient packaged recommendation is basically a myth; such people are really rare. In online conversations with people who volunteer for these projects, the only case I’ve really heard for them is that they might be helpful if your grandparents ask you how to vote, which is clearly a hopeful daydream.

It’s the dumb thing that smart people always do – assuming that the only reason other people haven’t done what you want is that you haven’t explained it to them yet. Unfortunately, politics doesn’t really work like that.

• Dan Davies is a former Bank of England economist and investment banking analyst