So we had another little foray out to look at a few bus stops. We deliberately chose to look at a small number, firstly to see what the error rate seems to be and secondly to check our system works.
After some short mailing list discussion I have decided not to amend the naptan: fields in the imported data, except that I delete the naptan:Verified=no field. I extracted a list of all of the nodes created from the changeset when they were imported and loaded this into a spreadsheet. I'll use this to record any anomalies I find. So far we have looked at 38 bus stops from the import, 2 are missing on the ground, 1 is on the wrong road, 3 have the wrong bearing, one is on the wrong side of the road, and 2 have the wrong code on the sign, the rest seem good, though I have moved a few too. Only another 1261 to check.
Our process works well now, but I could do with a render that shows bus stops at a lower zoom level to get more on a printed map. I add source=naptan + survey, I add shelter=yes if there is one and I will add layby=yes if there is one (none yet). If I find something wrong I add a note to explain the problem, and I write it into the spreadsheet. This works well for me because it looks as though we're the only people working on this in this area, but if anyone wants to join in I'll just share the spreadsheet online.
I have also contacted Hull City Council transport team to see if they are interested in feedback from what I find, I've yet to hear from them.