[OSM-talk] Réf.: Re: Proposal for import guidelines

[THEVENON Julien]

Le mar. 25 sept. 2012 21:22 HAEC, Jean-Marc Liotier a écrit :

>I like this proposal - from my very personal point of view it safeguards
>all the conflicting interests and reaffirms essential inflexible
>principles while cutting some slack to users who perform small local
>imports :
>
>The "bot=yes" tag identifies the import as such, to help moderators
>focus on that class of potentially widely damaging changes.
>
>"bot_url=<link to a page describing the automated edit" provides all the
>necessary context about the import, including quality and methodology
>issues specific to the source of data.
>
>The "bot_source_licence" tag clarifies the license status of the source
>at that point in time.
>
>The specific conditions (imports of more than a given number of nodes,
>continuously running scripts, edits affecting more than one country) for
>changesets for which a separate account is necessary are clear and
>non-equivocal, reaffirming the current requirement for a separate
>account while letting the users of occasional small-scale imports at a
>local level perform them with their personal account.
>
>The need to keep these conditions open-ended is a weakness that lets
>detractors claim that they are arbitrary, but I'm guessing that this is
>necessary to prevent users gaming the rules with stupid technical
>loopholes... Not quite transparent but practical.
>
>This proposal hits all the goals I have seen stated so far... Or are
>there others that are not satisfied by this proposal ?
>
>On the French list, some contributors are complaining that the
>changeset-level tagging makes the separate account requirement entirely
>obsolete. Technically, I believe they are right... But I hope they'll
>see that this proposal could be a fair meeting ground for an opportunity
>to improve the import process with better metadata and make it more
>flexible where necessary while not messing too much with the current
>international consensus

hi,

The proposal of changeset tags seems also good for me but I consider that the arbitrary criteria of node number is a weakness that fall in the trap of "stupid technical loopholes" mentionnedby Jean Marc.
it will be quite easy to split changesets to stay under the limit and avoid the use of a separated account. geographical area limitation (by example country scale) could also be hacked by splitting. that's why I personnally think that purely automatic script without manual refinement are better criteria to justify the use of a dedicated account unless there are some other reasons I'm not aware of that make the only use of tags unsufficiant

best regards
Julien