Friday, July 6, 2018

A Path to Extreme Longevity

Standard Disclaimer:  I'm a computer scientist, which of course also makes me an expert in fields of optics, holography, and cryobiology.

At a high level, I propose that the infrared holography technologies being developed by Dr. Mary Lou Jepsen at Openwater as a possible solution to the "unfreezing problem" being worked on by Alcor, a cryonics lab that suspends human heads with the hope of one day reanimating them.  This is, of course, only one small piece of the greater puzzle, and even after we record a full human head at the appropriate resolution, the process of digitally reanimating the head is still going to be a massive open problem, though tangible enough that it can be worked on at scale.

I believe this solution is ideal because the infrared holography scanning can be done in a completely noninvasive, nondestructive way.  It's great for Openwater because my understanding is their technology has a lot of trouble with movement, and the infrared holography has trouble with blood absorbing the scanning radiation.  Movement should not be a problem due to the freezing, and since Alcor replaces blood with medical grade antifreeze, that should vastly improve the potential depth of the infrared scans.

Milestone 1: Calling for Volunteers

A key difficulty for this process is finding a volunteer to be the first head to get scanned.  Alcor has a number of members signed up for cryopreservation, but for this project, volunteers would need to understand that in the very best case scenario, every thought, every memory, every detail of their life would be public domain, and picked over by countless scientists, potentially for generations to come.

Milestone 2: Developing the Scanning Process

A proper scanning harness would need to be developed to perform the most detailed scan possible, while also ensuring minimal damage to the subject.  The subject head(s) may need to be retrieved from cold storage and scanned regularly as scanning technology improves, so minimizing damage is essential to having progressively more detailed scans.

Milestone 3: Open Source a Head

Once the head scan is complete, the full scan data should be published as widely as possible.  I imagine this effort could be similar to this generation's Human Genome Project, with researchers around the world digging into the available data to make sense of it all.  This publishing of head data should ideally happen at some regular cadence, or any time there is a major imaging breakthrough.

Milestone 4: Memory Reconstruction

We have no idea how much precision is required to reconstruct a human memory, but we do know that they're in there somewhere.  Maybe the first memory researchers manage to extract will be along the lines of "Arizona is hot", but as the technology improves, researchers might be able to recreate the full scenery of memories as they were remembered.  The subject could even be asked to memorize some string of data, like a password, and researchers could be challenged to reconstruct that memory.

Milestone 5: Full Network Reconstruction

I imagine that eventually researchers will build a compiler to transform raw imagery into a representative neural structure.  It will be hard to know at first how much lossiness is acceptable to recreate full human cognition.

Milestone 6 : Reanimation

So, now what?  We have a massive neural network representing a full human consciousness, at some point we have to hit the "go" button and see what happens.  This will likely require an unimaginable amount of compute resources, and will likely at first run orders of magnitudes slower than typical human interactions, but what we learn from these experiences will likely inform subsequent generations of "brain compilers" and compute architectures.