Data and error bars

I spend most of my working hours as a scientist looking at data in research papers, and unpublished data from my group's experimental activity, and figuring out when these data are significant enough to represent a firm conclusion.

So I continue to be worried about the 48/52 result and it's strength. Let's call this ratio the "Brexit index". There are two problems.

(a) This was an index that had huge fluctuations (judging from polls) up to the vote; the vote captured one moment in time of this fluctuating volatile index. Now one does not embark on a project that takes at least two years, and has consequences for future decades, on the basis of one "read". One would not do this in buying a house, or some stocks... and even when we shop in the supermarket we know (for our regular products) the average price, so that we are not scammed too much by promotions.

(b) The second problem is that this "Brexit index" as well as being volatile due to people changing their minds is also subject to changes due to changes in the people eligible to vote. It is not easy for the human brain to contemplate very different timescales, and these days we are so bombarded with rapid changes and statements that I suspect we all focus (Brexit-wise) on the short term. Almost 1 million people die each year of old age, and 1 million people come into the electorate. Nobody can dispute that the vote was polarised between young and old (of course, not all the young and not all the old...) but for the sake of argument let me assume the case (there is the strong assumption here that on this issue the younger will not drift to the opinions of the older; I think this is a valid assumption because the desire to remain comes from personal experience and culture that was simply not shared by older people).
So, this is an index quite likely to drift towards remain. How quickly ? Well very roughly there is a 4% turnover of the electorate in less than 3 years.

In conclusion: (a) We don't know if the result has statistical significance; (b) it would be a huge paradox if the UK finally exited from the EU just when its electorate became in favour of remaining in.

What should we do? The brexit process will carry a huge "direct cost" in legal fees (there are reasonable estimates for this around), and much greater "indirect costs" that are at the moment impossible to estimate precisely because they depend on how the process goes. In this context, I suggest that measuring the "Brexit index" weekly, via a very well done poll (made public, and with such a sampling base that nobody can argue about its validity) is a necessity. In fact, this should be the first thing that the "Brexit Minister" should deliver. And then what ? (a) we will have an idea of the "error bars" on this result, and (b) we will be able to see what trends there might be on the index (obviously, demographics is just one aspect, and people might be responding differently as they understand better what is happening).

Finally, to state the obvious: if the "Brexit index" is not significantly < 0.5, and/or if there is a trend that sets it to get above 0.5 in the timescale of leaving the EU, then I don't see how there is any sort of mandate to leave.