I looked through my notes of the two weeks from Monday June 15 to Sunday June 28, 2020, and found almost nothing but work, work, and then, in a dusty, dark corner right at the bottom of my Emacs… more work.
A totally unexpected ramble on cycles, patterns and teleportation
Life seems to consist of the superposition of so many cycles over so many domains, each at a different frequency and phase.
At this moment, my backyard philosophy cycle seems to have taken a backseat, whilst my over-use of vehicle metaphors cycle can’t believe its luck.
That was the one where I talked about 98% of your atoms being recycled every year, which helps to understand why the concept of a single consistent “I” going through life is mostly an illusion.
Well, I mention this again right now, because these cycles do seem to be related, or maybe they are a contrasting idea.
On the one hand, maybe the I sitting here typing this has only 75% overlap with the I of 6 months ago.
Maybe the 25% that has now been “cycled” in (haha) has different priorities, and maybe over the coming months those priorities will shift the whole even more in a different direction.
In my ever-changing mind, all of this also casts interesting light on one of the favourite philosophical paradoxes of Star Trek nerds the world over: The Teletransportation paradox.
You might be surprised to know that Trekkies have bee wondering about this as far back as 1775! (see the linked Wikipedia article above)
In short (oh who am I kidding), this asks the question:
If it were possible to reconstruct you, atom-by-atom, in a different place (this is essentially what the Star Trek transporters do, with the one qualification that they mostly disassemble the departing you before assembling the arriving you), the you on the other side would have your thoughts and your feelings, but would it really be you?
If one were not inclined to thinking about matters such as these for entertainment, one might easily write it off as far too abstract.
However, how long do you think before we are able to copy (parts of) a human neural network to a different substrate that is able to run it?
In other words, how long do you think before we are able to upload a copy of a human brain to a sufficiently powerful computer?
AGI (Artificial General Intelligence) is certainly quite far off, but the discussions about it are quite serious already.
Parts of the technology required for AGI are at least in the same ballpark as some of the parts that are required for uploading.
(I’m reading another one of Neal Asher’s books at the moment. You couldn’t tell, could you?)
On the other hand (you’re going to have to scroll back to find the first hand. sorry), it could just be 100% boring human rhythm.
Just the other day, I got to bed far too late thanks to a coding session getting out of hand, followed by the above-mentioned book pulling me in again. Matthew Walker, I’m sorry.
The next day was hugely productive, with will-power to spare and my brain running on turbo mode until late in the afternoon.
On other occasions I pay close attention to sleep, diet, exercise, prayer beads, crystals, pyramids and aliens, and then the next day is a total wash: Brain checking out middle morning, and will-power calling in early to take the day off to play old Flash games online.
I called it human rhythm above, but I suspect an extremely influential part of that is mood.
Looking after the physical components (sleep, exercise, diet, pyramids) remains crucial, and does form the basis of one’s performance, but one’s mood can cause massive course changes regardless.
Trying to be a good human can be tough.
What I was actually planning to write
With the being “a bit stubborn” mentioned in the title, I meant that I was planning a perfunctory blog check-in despite not finding anything in my notes to write about.
(I consoled myself with the fact that at least I had the great photo above, contributed by occasional readers from right within my inner circle.)
As has happened a few times before, the ritual of sitting down (well lying in bed with laptop, sorry again Matthew Walker) to write a few sentences drew out the ramble above from the thoughts that just five minutes prior had indicated quite sternly that they had absolutely nothing to say.
Besides the photo above, I did want to mention the following two gems I ran into during the previous weeks:
Implicit Neural Representations with Periodic Activation Functions, a technique also known as SIREN for “sinusoidal representation networks”, uses sinusoidal activation functions instead of the ubiquitous ReLU to model images, sound, video and even signed distance fields.
This is a great example of implicit representation learning with neural networks, something that has been fascinating me for a while now.
In this case, they even demonstrate the neural network learning to predict a signed distance field (SDF) based on a point cloud with normals.
(At some point in the past, it looked like my new middle name was going to be “distance field”, based on the sheer number of times that it came up as the answer to any problem that we were trying to solve.)
The video by the authors is a great 10 minute summary of what you could do with SIREN:
More about John Carmack, programming god
I’ve written previously about the Carmack Productivity Method, named after a story about John Carmack using a CD Player as a tool to boost his already prodigious productivity and focus.
More recently, I really enjoyed reading the following write-up of the history of id software, including much more about John Carmack:
When he joined the group of young people who would later come to be known as id software, he was just 19 years old, but the programming genius within was already strong.
With Carmack at the core, id software would go on to write game programming history.
Thank you for reading any of this my peeps, I am truly grateful for the connection.
Please feel free to leave a comment, consider a nom de plume if you like!