[Tagging] How to tag Seveso sites ?

[Martin Koppenhoefer]

While I first was assuming this would comprise inactive but contaminated sites, I now see this is for operational sites only, which are dealing with chemical substances of which release into the environment could potentially pose a hazard to the people living nearby. Right?

How would we survey this? There is not much that can be surveyed worse ;-)
You would have to ask the operator of the site which are the substances they are dealing with (and they would not have to answer you, nor would it seem that if they answered their reply would be very reliable), then you would have to have the knowledge to understand their reply and to evaluate it, probably very few of us are in this position. So the only “reliable” source would be information collected, assessed and  released by the government. High risk could mean the substances are very dangerous, or a potential accident could release a lot of them, or it could mean the operations are executed in a way that an accident is more likely than it ideally should be, for example, if there wasn’t sufficient maintenance for a longer period of time, if the safety measurements aren’t the best/most recent available, problems in critical safety components have already been found, etc.

Am I guessing correctly that this is about kind and quantity of substances that are dealt with? 
The other possibly relevant information (for risk assessment) is likely not publicly available.

What is the benefit of having this information in OpenStreetMap? There is nothing the crowd could contribute to improve this data, we would be a mere distribution service of government data, and we would be at least as out of date as they are (because the process of verifying and inserting the data takes so time, as does the insertion in OpenStreetMap). 

If we still would decide to import this kind of assessment results, I would suggest to use keys which make it clear which criterion/system has been used (i.e. use specific keys for each source). While it may look similar on paper to a layman, it may be very different systems, or different practice of assigning risks even with the exact same system (but there’s no doubt that the actual systems/schemes are different).

Cheers Martin