You're right that 8 wards is too small to claim statistical validity, and the calibration was derived from those same 8 wards. That's overfitting until proven otherwise.
That's exactly why we are pre-registered predictions for 136 councils on May 7. I'll publically announce our predictions on May 1st.
If the calibration generalises, we have something. If it doesn't, we have a well-documented failure. I hope for the former, I can accept the latter :)
So to summarise the 2 potential questions... Is it any good? Does it work? The honest answer is: I don't know yet.
YouGov doesn't poll council elections. Nobody does. There's no commercial incentive to poll 5,000 individual wards at £10-50 per respondent.
That's the gap.
YouGov's MRP models predict at constituency level for general elections. We predict at council ward level for local elections.
Different product, different market.
On cost: 65,000 synthetic respondents cost us roughly £35 in GPU compute. A 1,000-person
YouGov poll costs £5,000-15,000. The accuracy isn't the same yet (we're at 75% winner accuracy on by-elections vs YouGov's 90%+ on generals), but we're predicting contests nobody else attempts.
The "in" is that local elections, by-elections, and ward-level prediction are completely unserved.
We're not replacing YouGov. We're filling the space below where polling is commercially viable.
And more than anything, we're testing and learning in public. If our panels get close to the real result after 7th May election, then there is something here... if not, well... I accept all your criticism :D
> YouGov doesn't poll council elections. Nobody does.
... Yes. They do.
> This is the smallest set of elections in the local authority cycle, but nonetheless throws up some intriguing contests. For the second year running, YouGov have used MRP to project the results of some key local authority battlegrounds.
You are right, I overstated that. YouGov did publish MRP projections for selected councils in 2024 and 2025.
The distinction I should have drawn is that MRP projects from national polling data down to local level using demographic models. It does not field ward-level samples. Our approach generates ward-level responses from synthetic personas with individual personality profiles.
Different method, overlapping output. Fair correction.
> Show HN: 65k AI voters predict UK local elections with 75% accuracy
Cambridge Analytica, is that you ?
a sample size of 8 after 9 rounds of overfitting?
You're right that 8 wards is too small to claim statistical validity, and the calibration was derived from those same 8 wards. That's overfitting until proven otherwise.
That's exactly why we are pre-registered predictions for 136 councils on May 7. I'll publically announce our predictions on May 1st.
If the calibration generalises, we have something. If it doesn't, we have a well-documented failure. I hope for the former, I can accept the latter :)
So to summarise the 2 potential questions... Is it any good? Does it work? The honest answer is: I don't know yet.
The paper says as much. May 8 will tell us.
The competition in this particular space is YouGov. Who have the same or higher accuracy rating, and are already being employed by all the incumbents.
This seems to get about the same results as a random sampling poll, but with AI instead of people. So higher infrastructure costs.
... Where's the "in", here?
YouGov doesn't poll council elections. Nobody does. There's no commercial incentive to poll 5,000 individual wards at £10-50 per respondent.
That's the gap.
YouGov's MRP models predict at constituency level for general elections. We predict at council ward level for local elections.
Different product, different market.
On cost: 65,000 synthetic respondents cost us roughly £35 in GPU compute. A 1,000-person
YouGov poll costs £5,000-15,000. The accuracy isn't the same yet (we're at 75% winner accuracy on by-elections vs YouGov's 90%+ on generals), but we're predicting contests nobody else attempts.
The "in" is that local elections, by-elections, and ward-level prediction are completely unserved.
We're not replacing YouGov. We're filling the space below where polling is commercially viable.
And more than anything, we're testing and learning in public. If our panels get close to the real result after 7th May election, then there is something here... if not, well... I accept all your criticism :D
> YouGov doesn't poll council elections. Nobody does.
... Yes. They do.
> This is the smallest set of elections in the local authority cycle, but nonetheless throws up some intriguing contests. For the second year running, YouGov have used MRP to project the results of some key local authority battlegrounds.
https://yougov.com/en-gb/articles/49288-local-elections-2024...
You are right, I overstated that. YouGov did publish MRP projections for selected councils in 2024 and 2025.
The distinction I should have drawn is that MRP projects from national polling data down to local level using demographic models. It does not field ward-level samples. Our approach generates ward-level responses from synthetic personas with individual personality profiles.
Different method, overlapping output. Fair correction.
I predict it with 50% without any data whatsoever. 75% does not impress me.
thats fair.