I see. I guess if there's a chance that we can stop genie from being out of box, we should try. You're also correct that tiered access can indeed mitigate issues. And it does seem like general models being as good and 'Bio Mythos' might genuinely not happen. So, exploring tightening access and data makes sense.
All of this requires co-ordinated action though as even one bad actor can significantly turn things for worse. And the fact that covid is fresh in policy makers' minds could help.
My points are more in the case of being or inevitably going to a point where we can 'reasonably freely' edit pathogens and vaccines. If we can do that, making it freer and restricting implementation is generally a better idea. And if possible to do safely, we should aim for that, I think. Because there is a lot of potential of good for everyone: humans, animals, other biological beings and maybe more (like plastic or rust eating bacteria for environment!).
I think we shouldn't restrict knowledge build up. That's like stopping knowledge build up for say material and other sciences that can be used in weapons and bombs. Imagine if we stopped doing that, it would have stopped so many good things we have now and can have.
In a sense, enough knowledge about how to build pretty terrible stuff is already public. But decent enough measures are in place for using it which prevents harm.
So, we should have measures that ensure that items that can make things usable by bad actors are restricted.
This also makes sure that we can also safely keep on improving knowledge and get better. Which brings me to the second point.
2. Offense is Sometimes Best Defense
We don't stop harm caused by computer malware and viruses by rolling back or stopping development and cutting off internet. We do it by patching things, being better at it, and moving on.
Now, I know that biosecurity has harsher security needs. But in a sense, the pandora's box is open. Unless you can stop everybody that has and is building BAIM or GAI, creating barriers for yourself is only increasing risk.
And besides, increasing capability also means that you've more room to monitor for threats and take care of it. Yes, it also means potential for bad is there and it needs to be accounted for (which restrict implementation if good enough should do); but the flip side is also true and shouldn't be buried in paranoia.
I see. I guess if there's a chance that we can stop genie from being out of box, we should try. You're also correct that tiered access can indeed mitigate issues. And it does seem like general models being as good and 'Bio Mythos' might genuinely not happen. So, exploring tightening access and data makes sense.
All of this requires co-ordinated action though as even one bad actor can significantly turn things for worse. And the fact that covid is fresh in policy makers' minds could help.
My points are more in the case of being or inevitably going to a point where we can 'reasonably freely' edit pathogens and vaccines. If we can do that, making it freer and restricting implementation is generally a better idea. And if possible to do safely, we should aim for that, I think. Because there is a lot of potential of good for everyone: humans, animals, other biological beings and maybe more (like plastic or rust eating bacteria for environment!).
I disagree with your framing - 2 reasons why:
1. Restrict Implementation, Not Knowledge
I think we shouldn't restrict knowledge build up. That's like stopping knowledge build up for say material and other sciences that can be used in weapons and bombs. Imagine if we stopped doing that, it would have stopped so many good things we have now and can have.
In a sense, enough knowledge about how to build pretty terrible stuff is already public. But decent enough measures are in place for using it which prevents harm.
So, we should have measures that ensure that items that can make things usable by bad actors are restricted.
This also makes sure that we can also safely keep on improving knowledge and get better. Which brings me to the second point.
2. Offense is Sometimes Best Defense
We don't stop harm caused by computer malware and viruses by rolling back or stopping development and cutting off internet. We do it by patching things, being better at it, and moving on.
Now, I know that biosecurity has harsher security needs. But in a sense, the pandora's box is open. Unless you can stop everybody that has and is building BAIM or GAI, creating barriers for yourself is only increasing risk.
And besides, increasing capability also means that you've more room to monitor for threats and take care of it. Yes, it also means potential for bad is there and it needs to be accounted for (which restrict implementation if good enough should do); but the flip side is also true and shouldn't be buried in paranoia.
Looking forward to see your response.