How do they account for this? a 0.0000001% chance of destroying 99.99% of current species for example? How do we measure how much we don't know about the envi

How do they account for this? a 0.0000001% chance of destroying 99.99% of current species for example? How do we measure how much we don't know about the environment? It seems to me we play pretty fast and loose when it comes to such things, often just because we want to spread new things as far and wide as fast as possible. I don't really understand how anything other than very careful and very slow introduction of such things can possibly be considered in any way safe. e.g. find a place you can try your new pesticide or whatever that you think will minimise the interaction with wider ecosystem. model the whole thing. run this for at least 50-100 years, if at any point it doesn't match your model, slam on the breaks til you can explain it (including it spreading more or less than you thought it would to other species and locations). It like we have inherited an app that runs our entire way of life that has billions of lines of code with no documentation but some are happy with applying patches without UAT, parallel run or backout procedure.