The separation between government and civil service is a vital one.
Governments come and go, but the civil service is permanent, and only works when it stands apart and acts as a bulwark against the worst excesses of politicians.
So a press release like that found on the Department of Work and Pensions website on Monday tips over an invisible but vital line of trust.
Titled 'Poll shows support for removal of spare room subsidy' the release publicises the results of an 'independent' poll which apparently shows that "there is strong public support for reducing under-occupation and overcrowding in social housing".
The DWP's use of the term 'spare room subsidy' is just about forgivable even as the rest of us call it a bedroom tax.
But the content of the release is both partial and misleading, relying on wafer-thin positive margins, picking and choosing only those that support its case and a wilful misunderstanding of results.
Ipsos MORI conducted the poll of over 2,000 people, and it is true that this is an 'independent' organisation as stated.
However, DWP funded the research and Iain Duncan Smith's fingerprints are all over it.
The introductory questions tell us plenty about the issues of perception that undermine the welfare state in the UK.
When asked at the beginning of the survey, 52% of people believe benefits are too generous, but only 36% said they knew a great deal or fair amount about the issue.
It is difficult to know how to respond to people who have strong opinions while professing not to know about a subject, but the survey - and DWP - should at least have highlighted this.
Assuming that the report follows the same order as the questions, it looks like it was constructed to lead people in the direction IDS desires.
By starting with linking the bedroom tax to the overall notion of 'benefits' - a term quite separate from specific social support payments as we have written before and one that has become entirely debased by the media and government - respondents are invited to think of the bedroom tax as part of the solution to an invented problem, namely that the benefit bill is unaffordable and is primarily an issue of unemployment.
This isn't borne out by the figures; if the £200 billion total is too high, why has the government 'triple-locked' the largest portion of it, pensions, and committed to increasing them by more than the rate of inflation?
The skilful design of the questions appears to have played a big part in the results. Only five weeks ago a Sunday People poll showed that 60% of people wanted the tax scrapped, and this figure has been growing rapidly across the year.
The DWP appears to have taken the most partisan possible reading of the research it has funded.
Using a press release to announce 'strong support' for the bedroom tax when fewer than half those surveyed actually said they supported it 'in principle', surely the most open version of the question, looks dodgy to say the least.
But when the report digs further into the fairness issue - absolutely key to understanding the bedroom tax - the findings do not make pretty reading for either the government or DWP.
It states that "if RSRS [removal of spare room subsidy] encourages those affected to take up employment (52% support, 20% oppose), while twice as many support the policy than oppose it if it means those affected have to move to find more affordable accommodation in the same area (49% support, 24% oppose).
However, more people oppose RSRS than support it if it means that those affected have less income to cover living costs (31% support, 35% oppose), and opposition grows further if RSRS means that those affected have to move to find more affordable accommodation in a different area (31% support, 40% oppose)."
So people support the bedroom tax if it encourages people to get into work (in some areas up to 55 people chase each job, so it isn't likely to help much here) and if people can find alternative accommodation in their area (they can't - in Merseyside only 155 of 26,500 households were able to downsize to avoid it), and don't if they have less income (they do, and are getting into debt as a result) or have to move to a different area (they might, but there is a shortage of smaller affordable homes so 96% can't).
So the headline on the press release that the majority support the tax isn't borne out by the results of the polling even given its leading design and poor interpretation.
The DWP comes out of this badly and needs to separate its own role and interests from those of the coalition quickly, or it will lose all credibility.
Publishing - and publicising - all the polling and other ratings it commissions and collects would be a good start to the healing process.
Last week it trumpeted a 98% user satisfaction rating for its online births, deaths and marriages registration service, and UnemployedNet called for it to publish these as a matter of course as part of its commitment to open government, rather than picking and choosing only positives as a PR exercise.
That has to include ratings for jobcentres, the work programme and other government schemes which are horribly unpopular with those forced to participate in them.
These may not make the kind of headlines the coalition wants, but the DWP could make a huge contribution to driving the kind of improvements these services so desperately need.