I can never decide whether I have a terrible memory or whether I just place a very high premium on maximizing my working memory. But, either way, I try to keep as much "to do" sort of knowledge out of my head as possible.
One of the main ways I do this is by ensuring everything goes in a logical place. I don't usually have to remember where I put things because I just ask myself, "Where would past me have put this?" and the answer is usually where it is.
Alas, I have a wife with ADHD, so my house is sort of like living in an Etch-a-Sketch where objects are randomly relocated when I'm not looking.
I’m one of those ADHD spouses and I WISH I got a consistent answer from my brain when I ask “where would I have put it”. Sometimes it’s because I temporarily put it down and then forgot it existed, but the rest of the time my brain is on a choose-your-own-adventure for organizational questions. At least, deep down, it annoys me just as much as my wife.
Most people would likely curl into a ball and cry if they had to interact with my computer desktop or file system organization, try as I might…
I frequently place my car keys inside or under a thing I need to take with me when I leave. Costs me having to search for my keys, but only in the times when I would have otherwise forgotten the thing.
> When I need to remember to throw trash away, my wife or I put the bags right at the foot of the front door.
Not so obvious but difficult. As someone whose partner has severe ADHD, I can assure you that if the author wants to imply that no memory capacity is necessary to make this simple construct work, he is wrong in my opinion. It takes a “shallower” (?) form of memory to link something like this (you can definitely link it -- I do it similarly - I call it routine)
> To keep track of how many hours I've worked in a day, I move Lego bricks from one side of my computer's monitor to the other at every periodic break.
I think it's clearer in this case. The whole thing only works because you have memorized that the lego bricks on one side of the work table mean a certain thing.
In the end, it's a bit about rewiring your neurons because the original wiring didn't work well enough under real conditions.
when i want to make lasting changes in my life i now think about IF -> THEN triggers and "wire" them. Since I'm very principle driven (a bit on the spectrum) this makes it easy for me to maintain routines (but also problematic when I can't for $reasons).
I really liked the post and dont want to badmouth anything.
Once a week I take the trash (in a bag) and recycling (in a bin) from our kitchen out to the larger bins outside our garage, wheel the larger bins out to the street, and then go back inside. I had chronic issues forgetting to bring the recycling bin back inside, so I had the clever idea of putting the bin directly blocking my path back inside after I empty it into the larger outside bin.
This helps only slightly. I still regularly forget the bin on my way back inside as I automatically pick it up and move it to the side without thinking about what I’m moving or why I put it there.
Slightly tangential, but stigmergy is a fascinating topic. If anybody recalls Ant Colony Optimization from a couple of decades ago, it was based on this: https://en.wikipedia.org/wiki/Ant_colony_optimization_algori... -- I was researching it for a bit, and while the reality never matched the hype, people actually shipped a few related projects in the areas of logistics. It eventually went nowhere, I suspect as computers became more powerful and other, less heuristics-based approaches worked better.
I still think stigmergy is under-appreciated, especially in the context of decentralized systems. The main reason centralization is such a powerful force is because you can do so, so much more with data aggregated in one place (e.g. discovery, spam filtering, and "big data" in general.) But if we want to achieve similar capabilities at similar scale in fully decentralized systems, stigmergy might be a powerful tool.
It's interesting that we've gotten away from this as more and more things have moved onto computers.
I worked with some legacy systems in the government that still used paper case files and it struck me one day how the "status" of a case was, in a very real way, represented by the file's physical location. Just submitted? The file is in the mail room. Ready for processing? In a clerk's inbox. Ready for review? In a supervisor's inbox.
It's basically a decentralized system, where each stage knows where the file goes next. Much different than our centralized databases today.
"People seem to do this all the time without much thought:"
I do it all the time with that because I know otherwise leads to catastrophe.
I want set a tell a woman I was going to be late for a date because I couldn't find my car keys. For some unknown reason I'd put them in the grocery bag and put the grocery bag directly into the refrigerator. It took me hours to find--hours I hadn't planned on because I usually put my keys in the same spot to avoid the nearly certain trouble that happens when I assume my future self will just remember what I did.
Along similar lines, after decades of struggling with the demands of multi-taskng, I came to realise that I am clearly incapable of it. I have now learned to work like Javascript. I have a list of items lined up, which is processed sequentially, in a single thread. Anything new that is not pressing goes to the bottom of the list. Urgent inescapable interruptions go to the top.
So - physical memory? I've been thinking recently about how we already live in an advanced computing environment: the world.
Our digital realm maybe mixes map for territory. And the point here is to make marks on the territory and throw the map away. Kind of makes me think of Lucy Suchman talking about navigating as situated action rather than planful analysis.
I've always thought that the economy works this way - we coordinate on a mass scale and on a decentralized basis. Everytime you buy or sell something, you send a price signal which is heard around the world.
To me it seems like LLMs are basically memory for humans as a whole. By interfacing with them, you can extract the knowledge, eliminating the need to remember things.
This has been the case for a while with search engines. I'm convinced our brains have evolved (atrophied?) to avoid having to remember things that you can simply look up on your phone in a matter of seconds.
Looking at this through transhumanist metaphors, like in any of dozens of scifi works to cover future cognitions well (my favourite is Eclipse Phase, if you want a recommendation).
Exocortices would be great if the modern technology was there to support it, but the current paradigms make memory-access too slow and unreliable to be worth the agency benefits, especially since current brain formats need internally-stored memories for default-mode idea generation in the background and can't make good use of location pointers for dreams.
I love this article.
I can never decide whether I have a terrible memory or whether I just place a very high premium on maximizing my working memory. But, either way, I try to keep as much "to do" sort of knowledge out of my head as possible.
One of the main ways I do this is by ensuring everything goes in a logical place. I don't usually have to remember where I put things because I just ask myself, "Where would past me have put this?" and the answer is usually where it is.
Alas, I have a wife with ADHD, so my house is sort of like living in an Etch-a-Sketch where objects are randomly relocated when I'm not looking.
I’m one of those ADHD spouses and I WISH I got a consistent answer from my brain when I ask “where would I have put it”. Sometimes it’s because I temporarily put it down and then forgot it existed, but the rest of the time my brain is on a choose-your-own-adventure for organizational questions. At least, deep down, it annoys me just as much as my wife.
Most people would likely curl into a ball and cry if they had to interact with my computer desktop or file system organization, try as I might…
I always tell my wife that she needs to give me a trigger warning before I see all the unread emails in her Gmail inbox.
SDAM (Severely Deficient Autobiographical Memory) is awful; we tend to just have blocks of a few years of memories.
The only benefit is you can't be gas-lit intentionally because you are always gas-lit, so you believe nothing. :-)
I frequently place my car keys inside or under a thing I need to take with me when I leave. Costs me having to search for my keys, but only in the times when I would have otherwise forgotten the thing.
Very interesting.
I may take it to literal.
> When I need to remember to throw trash away, my wife or I put the bags right at the foot of the front door.
Not so obvious but difficult. As someone whose partner has severe ADHD, I can assure you that if the author wants to imply that no memory capacity is necessary to make this simple construct work, he is wrong in my opinion. It takes a “shallower” (?) form of memory to link something like this (you can definitely link it -- I do it similarly - I call it routine)
> To keep track of how many hours I've worked in a day, I move Lego bricks from one side of my computer's monitor to the other at every periodic break.
I think it's clearer in this case. The whole thing only works because you have memorized that the lego bricks on one side of the work table mean a certain thing.
In the end, it's a bit about rewiring your neurons because the original wiring didn't work well enough under real conditions.
when i want to make lasting changes in my life i now think about IF -> THEN triggers and "wire" them. Since I'm very principle driven (a bit on the spectrum) this makes it easy for me to maintain routines (but also problematic when I can't for $reasons).
I really liked the post and dont want to badmouth anything.
Once a week I take the trash (in a bag) and recycling (in a bin) from our kitchen out to the larger bins outside our garage, wheel the larger bins out to the street, and then go back inside. I had chronic issues forgetting to bring the recycling bin back inside, so I had the clever idea of putting the bin directly blocking my path back inside after I empty it into the larger outside bin.
This helps only slightly. I still regularly forget the bin on my way back inside as I automatically pick it up and move it to the side without thinking about what I’m moving or why I put it there.
Can be a fun project to automate with homeassistant.
Slightly tangential, but stigmergy is a fascinating topic. If anybody recalls Ant Colony Optimization from a couple of decades ago, it was based on this: https://en.wikipedia.org/wiki/Ant_colony_optimization_algori... -- I was researching it for a bit, and while the reality never matched the hype, people actually shipped a few related projects in the areas of logistics. It eventually went nowhere, I suspect as computers became more powerful and other, less heuristics-based approaches worked better.
I still think stigmergy is under-appreciated, especially in the context of decentralized systems. The main reason centralization is such a powerful force is because you can do so, so much more with data aggregated in one place (e.g. discovery, spam filtering, and "big data" in general.) But if we want to achieve similar capabilities at similar scale in fully decentralized systems, stigmergy might be a powerful tool.
It's interesting that we've gotten away from this as more and more things have moved onto computers.
I worked with some legacy systems in the government that still used paper case files and it struck me one day how the "status" of a case was, in a very real way, represented by the file's physical location. Just submitted? The file is in the mail room. Ready for processing? In a clerk's inbox. Ready for review? In a supervisor's inbox.
It's basically a decentralized system, where each stage knows where the file goes next. Much different than our centralized databases today.
"People seem to do this all the time without much thought:"
I do it all the time with that because I know otherwise leads to catastrophe.
I want set a tell a woman I was going to be late for a date because I couldn't find my car keys. For some unknown reason I'd put them in the grocery bag and put the grocery bag directly into the refrigerator. It took me hours to find--hours I hadn't planned on because I usually put my keys in the same spot to avoid the nearly certain trouble that happens when I assume my future self will just remember what I did.
Insightful and pithy! Well written too...
Along similar lines, after decades of struggling with the demands of multi-taskng, I came to realise that I am clearly incapable of it. I have now learned to work like Javascript. I have a list of items lined up, which is processed sequentially, in a single thread. Anything new that is not pressing goes to the bottom of the list. Urgent inescapable interruptions go to the top.
So - physical memory? I've been thinking recently about how we already live in an advanced computing environment: the world.
Our digital realm maybe mixes map for territory. And the point here is to make marks on the territory and throw the map away. Kind of makes me think of Lucy Suchman talking about navigating as situated action rather than planful analysis.
I've always thought that the economy works this way - we coordinate on a mass scale and on a decentralized basis. Everytime you buy or sell something, you send a price signal which is heard around the world.
The housing crisis was an ant mill? https://en.wikipedia.org/wiki/Ant_mill inb4: breadcrumb dollars, a bit coin, if you will https://www.designboom.com/wp-content/uploads/2022/05/bit-co...
To me it seems like LLMs are basically memory for humans as a whole. By interfacing with them, you can extract the knowledge, eliminating the need to remember things.
And become the perfect puppet for the ruling class! 1984's got nothing on us.
This has been the case for a while with search engines. I'm convinced our brains have evolved (atrophied?) to avoid having to remember things that you can simply look up on your phone in a matter of seconds.
Looking at this through transhumanist metaphors, like in any of dozens of scifi works to cover future cognitions well (my favourite is Eclipse Phase, if you want a recommendation).
Exocortices would be great if the modern technology was there to support it, but the current paradigms make memory-access too slow and unreliable to be worth the agency benefits, especially since current brain formats need internally-stored memories for default-mode idea generation in the background and can't make good use of location pointers for dreams.