tl;dr: science is in the eye of the beholder, you can only know if it’s science if the methods are transparent and you have access to data, as well as critiques from unbiased parties.
This thread seems to have formed two sides:
unless it’s published, peer reviewed and replicated it’s not science, and
LeCun is being elitist, science doesn’t have to be published. This point of view often is accompanied by something about academic publishing being inaccessible or about corporate/private/closed science still being science.
I would say that “closed”/unpublished science may be science, but since peer review and replication of results are the only way we can tell if something is legitimate science, the problem is that we simply can’t know until a third party (or preferably, many third parties) have reviewed it.
There are a lot of forms that review can take. The most thorough is to release it to the world and let anyone read and review it, and so it and the opinions of others with expertise in the subject are also public. Anyone can read both the publications and response, do their own criticism, and know whether it is science.
If “closed” science has been heavily reviewed and critiqued internally, by as unbiased a party as possible, then whoever has access to the work and critique can know it’s science, but the scientific community and the general public will never be able to be sure.
The points folks have made about individuals working in secret making progress actually support this; I’ll use Oppenheimer as an example.
In the 40s, no one outside the Manhattan project knew how nuclear bombs were made. Sure, they exploded, but no one outside that small group knew if the reasoning behind why they exploded was correct.
Now, through released records, we know what the supporting theory was, and how it was tested. We also know that subsequent work based on that theory (H-bomb development, etc.) and replication (countries other than the US figuring out how to make nukes, in some cases without access to US documents on how it was originally done) was successful and supported the original explanations of why it worked. So now we all know that it was science.
If we put Elon Musk in a box along with a detector calibrated to detect the emission of a radio active particle, with a device that will cause Elon Musk to do science if it detects the particle and make up bullshit if it doesn’t, does Elon Musk remain an arrogant asshole no matter what the particle does?
tl;dr: science is in the eye of the beholder, you can only know if it’s science if the methods are transparent and you have access to data, as well as critiques from unbiased parties.
This thread seems to have formed two sides:
I would say that “closed”/unpublished science may be science, but since peer review and replication of results are the only way we can tell if something is legitimate science, the problem is that we simply can’t know until a third party (or preferably, many third parties) have reviewed it.
There are a lot of forms that review can take. The most thorough is to release it to the world and let anyone read and review it, and so it and the opinions of others with expertise in the subject are also public. Anyone can read both the publications and response, do their own criticism, and know whether it is science.
If “closed” science has been heavily reviewed and critiqued internally, by as unbiased a party as possible, then whoever has access to the work and critique can know it’s science, but the scientific community and the general public will never be able to be sure.
The points folks have made about individuals working in secret making progress actually support this; I’ll use Oppenheimer as an example.
In the 40s, no one outside the Manhattan project knew how nuclear bombs were made. Sure, they exploded, but no one outside that small group knew if the reasoning behind why they exploded was correct.
Now, through released records, we know what the supporting theory was, and how it was tested. We also know that subsequent work based on that theory (H-bomb development, etc.) and replication (countries other than the US figuring out how to make nukes, in some cases without access to US documents on how it was originally done) was successful and supported the original explanations of why it worked. So now we all know that it was science.
If we put Elon Musk in a box along with a detector calibrated to detect the emission of a radio active particle, with a device that will cause Elon Musk to do science if it detects the particle and make up bullshit if it doesn’t, does Elon Musk remain an arrogant asshole no matter what the particle does?
Trick question – Elon Musk being anything but an arrogant asshole is impossible under the Standard Model.