Yes, they don't make it easy to follow what they're up to do, do they?
Fit to Work decisions.
These are made by a two-stage process. The ATOS tester (a doctor, nurse or physiotherapist, randomly assigned except for people with neuro probs) fills in a multiple-choice computer questionnaire which awards points for inability to do different activities. There are spaces for tester comments, but these are often not used and when they are often contain comments on entirely different testees (my last one had a reference to me using a mobile phone during the interview and handling a packet of tablets - neither happened).
The test and score then go to a back office of the DWP where clerks called Decision Makers who have no medical knowledge look at the number of points and if it is over 15 award ESA, if it is less than 15 do not award ESA.
This creates a nice blame-boomerang. The ATOS tester is only filling in their questionnaire, it's not their fault what the Decision Maker does. The Decision Maker is only going on the number of points, they can't be expected to interpret medical reports.
Information does not seem to be shared with the separate clerks in the JobCentre part of the DWP. I was called for a JobCentre Work-Focussed Interview while awaiting proper classification - my test above having been set aside by ATOS as not fit for purpose. The WFI bloke ("Disability Advisor") and his JobCentre boss were most surprised when I informed them I was still awaiting testing. At one stage they were asking me to have a correspondence with the Decision Makers and "copy them in" on it, to demonstrate I was telling the truth. I pointed out that they were the DWP, the Decision Maker was the DWP, perhaps they could ask their own colleague directly, especially since the basic premise of their request was that they wouldn't believe info when it came from me. There was very clearly no Data Protection issue stopping them doing this as the data was still within the DWP and anyway the ESA application form requires you to sign permission for your data to be shared with the postman a large number of agencies plus any we think of later.
By the way, two specific issues about the ATOS test give the game away that it is not "in good faith".
The ESA form asks for permission to approach your GP/others for medical reports, and they used to send the GP a standard form - I've sat in the surgery while the GP filled it in. The requirement to seek these medical reports has now been removed. I was completely unaware the last time I was tested that ATOS hadn't approached my GP, and I was therefore not in a position to independently supply a GP's report. But appeal cases make it clear the testers often don't read reports even when they are supplied.
The second thing is the way some of the questionnaire scoring has been changed. If you're interested I'll dig out the details, but in brief, the question about going up and down stairs used to attract different points for 4 or 5 different levels of capacity: cannot go up one step, can go up but must rest halfway, etc These gradations have been stripped out and the Qs are now (iirc) 1) Can you go up two steps? 2) Can you not go up steps at all? So people who can climb steps but slowly, with pain and needing to rest half-way will now be graded the same as a mountain climber. This was not an inadvertent oversight; it was a specific change from an existing test. Someone sat down and decided to do this.