A Bunch of Ideas from a Crazy Dad Who Woke up at 3am
If I don’t write them down now, I won’t have time later
I’m someone who generally prefers a good three hour chunk of time to relate an idea. That’s mostly because my brain seems optimized for association and if I explained ideas to people the way that I conceived them, it just wouldn’t make any sense. This necessitates me spending a lot of time going back and forth with myself trying to answer the question “How would I explain this if I were a normal person?” which I will define more rigorously below.
Since I no longer have three hour chunks of time and likely will not for several years, I’m just going to do a scatter shot of things that I keep in the back of my head that are probably better for the world to know in case I get hit by a bus or something.
A good way to define sanity/insanity and how this is probably useful for AI
As I’ve referenced here multiple times I consider myself to be kind of nuts. One of the most useful things I ever did when I was at my most nuts was go to a support group for people who had survived childhood sexual abuse. It really led me to understand what was different about me because of what I had endured —I very strongly dislike to have my back to doors, and neither did anyone else there, which made for some interesting seating arrangements— as opposed to what was unique to my own personality —I spend a lot of time in my head thinking through overly philosophical what-if scenarios.
One of the things I really started to put together there was a systemic way of thinking about how crazy people are in terms of outward behavioral effects. A lot of the way we define things around mental health lately seem to have to do with internal states, which are always unfalsifiable, and just very much limit the utility of those observations. When you say you feel like a certain something, it doesn’t necessarily tell me how that is going to express itself in your outward behavior in the world of atoms. Defining things this way is almost always a bad idea. This is also my biggest beef with the way that Sam Harris defines Free Will —which if extended to cars means that because a car has both a motor and also wheels, separately, that it cannot actually drive. That obviously doesn’t work.— If you’re interested in my own view on free will you can look up Compatabilism on Wikipedia.
Being in that room was one of the only times in my life I noticed that I was generally speaking “more sane” than most of the other people there. Maybe it’s because my abuser only got me once, and most of my trauma came from survivor’s guilt that I was unable to speak up in defense of my cousins, but I determined that was an unanswerable and almost uninteresting question. What mattered was that day to day, I could pretend to be sane enough that most people would never know about the weird terrors I was constantly navigating, and at least several people there could not. I was also able to still do a very good job imagining what sane behavior would look like, and at least several people there could no longer able do a very good job at that.
This led me to the following two scales
Sane Simulation
Type 1: You can convince another person that you are sane without having to tell them anything you believe to be a lie.
Type 2: You can convince another person that you are sane, but you have to tell them things that you believe are untrue.
Type 3: You can hold a conversation with someone, in the sense that what they say to you changes what you intended to say back to them and vice versa, but you are unable to determine which lies would be appropriate to tell them to convince them that you are sane.
Type 4: You can no longer hold conversations using any medium. What a person says to you has no impact on what you say back to them.
Please note that this doesn’t exactly overlap with what is objectively true. You can absolutely know something that everyone else will later know to be correct and still appear to be completely insane when you relay it to them. In terms of effect, and the world of atoms, sanity is nothing more than the ability to understand what will cause another person to label you as insane and avoiding those behaviors.
I’m somewhere between one and two on this scale, right along with most people. It’s probably not good to be a one. However, I’m probably a lot closer to two than baseline.
Sane Effort
Type A: Your default state is to appear sane. It takes no effort for you to maintain the appearance of sanity around others. No matter how long you are around someone else they would be convinced that you are sane.
Type B: You can maintain the appearance of sanity long enough for you to never run out of energy to appear as insane to another person. You can always convince someone you are sane even if it takes some effort.
Type C: You can generally maintain the appearance of sanity, but occasionally it becomes too effortful to keep up. Over a long enough period, another person would define you as insane.
Type D: You cannot summon the will power to appear as sane to people even though you can imagine a sane response.
Type D: You cannot summon the will power to appear as sane to people and you cannot imagine a sane response.
I’m a solid type C here. A much better C than I was in the past as it will generally be years or months between “freak out” events but I definitely have to manage the way I appear to other people on a daily basis. For anyone reading with PTSD, therapy and exposure really do help, as painful as both can be.
I move these definitions around every so often but you’ll see it’s about in-groups and out-groups. This is something you could actually build a training set around for an AI, and part of the reason why I think something like the Forum and the Index would be very valuable. It would be a training set for sanity. We want to train an AI to give us answers that we will generally find to be sane and have it evolve over time to keep it in lock-step with what humans define to be correct.
There is no one singular awesome way to interpret reality. Even if we break it down to a single equation there is a level of subjectivity in the interpretation. That’s where I get a bit annoyed with people. You can’t have the one right answer. You can only have an interpretation of the underlying reality that’s in line with human norms in a way humans can accept.
This gets a bit complicated when you have schisms between constellations of people on what is sane and what is not insane and there are things you can do there to mediate and make-peace but I’m already running low on time so I can’t give my solution to the problem of war here but it involves people being able to move a little bit toward type 2/B in both categories to find a workable agreement.
The Super Ego-rithm
I wish that Apple and Microsoft would give me the ability to lock myself out of certain functionality in the following manner:
Make my super ego more powerful than my hedonic reward system.
Here’s a real life example.
I discover that I am browsing Twitter too much.
I know this is detrimental to my mental health. I know I could be using the time more productively and that this causes me to not be fully focused on my work. But the time preference in my brain doesn’t match between systems. And the hedonic system wins every time.
I want to be able to go into my devices and say “don’t let me do this” and have them block me from accessing Twitter. I want this to literally be a law that works for every device. I can’t even use any kind of computer without this being in the interface somewhere, tied to my unique identity. I’ll explain why in a bit, yes, but this has to sound Orwellian for now. Stick with me.
I can only change this functionality if, over multiple days, I take the consistent action to request the update. Any deviation from making the request over that span will cause a reset of the timer logic. So, say for thirty days I have to consistently want to be on Twitter enough that I go in and ask for this for thirty days in a row. Without ever once being interrupted by my super ego saying “Bro, you know what that does to you” and getting me into a car to go on a hike at a state park instead.
I am made up of multiple systems. One of these systems is very good at convincing the other systems that they should do things like go look at Twitter all day to see what the latest drama is. The other systems don’t seem to be able to stop this from happening even though they know, and could summon the effort to tell that system “hey stop” a single time, they can’t do it consistently. Except the opt-ins are now all aligned to my hedonic system.
I want to give my super ego a tyrannical ability to rule over my hedonic system. I only have to summon the will to do the right thing once in a given period of time and my hedonic system can no longer make actionable changes to my life. Give more weight to the part of me that wants to make the eighty year old version of myself happy.
I even want money to work this way. I would like to be able to force myself to be unable to buy ice cream. I fucking love ice cream.
As the future gets better and better this becomes more and more important. What happens when there are sex robots? I consider this to be an abomination, by the way, and think it should probably be illegal because the most dangerous thing for humans to do is to erase the problems to which we arose as the answer, but let’s say I can’t get them banned for everyone. I know I shouldn’t ever buy one but what if I get an ad that says they’ll make it look like an aerobics instructor from a show I used to watch when I was thirteen in the 1990’s? The one that caused my sexual awakening by doing bullshit exercises that looking back, didn’t seem to increase health a lot and were pretty much designed to awaken the sexual desire of a thirteen year old? How could I even say no to that?
Unless I gave my super ego the power to say no once, and set my hedonic system in a state where it had no power to override my super ego.
So, I want there to be a Big Brother just not a guy in a weird suit at Davos. I think this would be super important for a lot of people with a porn or any other kind of addiction. I want Big Brother to be the part of myself that thinks things like “Maybe I don’t need another Oreo Blizzard.”
I’m way more of an asshole to myself than any weirdo in a suit at Davos could ever be.
Sanity thing - call it putting on a human suit. Find it more exhausting with each passing day. In person is easier. Phone is harder.
May sex bots damage the first adopters’ gonads and curse the industry with uninsurability.
Forgot to add: the super ego-rithm but for neural implants.