[OSM-talk] [OSM-dev] Developers requested to help provide"completeness" tools

[Chris Morley]

Andy Robinson (blackadder-lists) wrote:
> Verification is a whole new ballgame.

I think throughout this discussion there is tendency to get hung up on 
the word "complete", which has been used as a shorthand but is being 
interpreted differently. In everyday use it has an implication of 
perfection and that there is nothing more to be said.

I think what we should be talking about is an area being "filled in", 
without implying perfection or immutability. You should expect as high a 
proportion of mistakes in a filled-in area as in an incomplete one, but 
fewer omissions. However, if there is a blank space on the map, you can 
assume that it really is empty in a filled-in area, but you would not 
know if it was in an incomplete area.

Measures of quality and guarantees of correctness require filled-inness, 
but I think should be regarded as more advanced concepts. I agree with 
Andy, we should walk before we run - start with an implemention of 
filled-inness - verification, etc. can come later.

Chris