Aspiring Rationalist Blogger and Vlogger, self-identifies as a sentient starship and temporarily embarrassed superintelligence.
As long as we can include a link back to the original blog post, sure.
I was considering something like a bike 'assistance'
A bike is a form of mechanical leverage assistance that allows us to reach speeds we would not otherwise be able to with the human body. Similarly, hang glider wings and the like are a form of mechanical leverage that allows us to fly when we would not otherwise be able to with the human body. Humans are kind of pathetic without technology, we can't do so much as push a nail into a piece of wood without some form of mechanical assistance.
I can see the basis of arguments that ethics could not be solved with hard science, I disagree with them but they at least have some basis. But psychology? Really? Are human beings not part of reality? Are human brains just magical boxes beyond our mortal comprehension? The hard problems of consciousness will be solved eventually. Cognitive neuroscience is making strong strides. Once we have a map of the connectome we'll be well on our way to really understanding how brains work. Psychology should absolutely be treated as a hard science.
" it's obvious that designing a society also involves solving questions outside the hard sciences"
these questions should not be outside the hard sciences, that's the point Alfred Korzybski was making all the way back in 1921. There's no reason we shouldn't be trying to treat ethics and psychology like hard sciences.
I'm actually working on a post for that, but writing it has been rather hard.
Not since I've updated around keeping my identity small. I intend to but my writing queue is quite long at this point.
These are sufficiently good questions that I want to research them and I'll get back to you.
I think what I'm describing here is a bit more advanced in terms of internal rearrangement than "simple mental parts"
There's two ways to do tulpas. There's the right way, and the way most people do it.
The right way is to do it from a place of noself/keeping your identity small. Don't treat your tulpa like a separate person any more than you would treat your internal sense of self like a separate person. Treat them like a handle for manipulating and interacting with a particular module/thought structure/part of your mind, taking unconscious and automatic things and shining a bit of Sys2 light on them. Basically using the tulpa as a label for a particular thought structure that either already exists, or that you want to exist in your head, allowing you to think about it in a manner that is more conscious and less automatic.
Doing this correctly gives you a greater degree of write-access to various semiconscious/subconscious parts of your head and makes it easier to retrain automatic response patterns. This could be considered in the same vein as how Harry uses his various house characters in HPMOR, although he just scrapes the top level with them and doesn't use them to really change himself in useful ways like he could potentially be doing. This way is also harder than the way most people do tulpamancy because it requires ripping apart and rebuilding your conception of your original self with a goal for greater functionality. It also requires keeping your identity small and internalizing the idea of noself in a way that most people don't want to do.
Then there's the way most people do tulpamancy, which is to build the tulpa out of identity and treat it like an entirely separate person who "lives in your head with you" and who has an equal say in decisions as "you." From the perspective of having internalized noself/keep your identity small, this is exactly as dumb as it sounds and looks. "Hey what if you destroyed your self control by handing it off to a random agent that you fabricate" or "Hey what if you created an internal narrative where you're powerless in your own head and your self is forced to argue and compete and try to negotiate with some other random self for processing time and mental real estate?"
Some people say tulpas can learn other skills and be better at them. Others say they've never lost an argument with their tulpa. Tulpas can be evil. Tulpas are slavish pawns. Tulpas can take over your body, tulpas never take over bodies. Tulpas can do homework. Tulpas can't do math.
Most people do identity-style tulpamancy, and that's where all this contradictory and at times really messed up behavior comes from.
An out-of-control dark rationalist tulpa that fights me for mental and physical control sound absolutely terrifying.
Right, so how does this happen? It happens because there are narrative layers that you're using (right now) to define what you can do in your own head, and that narrative layer is the thing being modified by tulpamancy. The problem is most people don't consciously try to modify that layer, they assume the way that layer works is some objective fact, argue about its properties with other tulpamancers online, and don't think about trying to change it.
The more power they handoff from their conscious mind to that narrative layer, the more "independent" the tulpa will seem at the cost of making the original self increasingly powerless within their own mind.
Intentionally grabbing hold of that narrative layer and modifying that such that stuff like multiple selves and the like are simply downstream results of upstream modifications will result in a much more cooperative and internally stable system since you can define the stability and the interactions as part of the design instead of just letting some unconscious process do it for you.
So basically, tulpamancy can be useful and result in greater functionality and agency, but only if done from a place of noself and keeping your identity small. If you haven't worked on grinding noself and keeping your identity small, that should definitely be the first thing you do. When finished, then possibly return to tulpamancy if you still feel like there's room to improve with it.
Did I actually do the right thing here? I honestly couldn’t tell you. There’s certainly an argument that could be made that I didn’t fully think through the consequences of my actions or what effect they would have on me. There’s also an argument that could be made that my defiance was rather pointless since the olive bar is still open, and if I was going to do something that crazy, I should have saved it for when I knew it would make a difference.
The problem is that barring near-omniscience you can’t really know when that will be, all you can do is play your hand and let the cards fall where they may. Would I have still tried to do this knowing everything I do now? Probably not. Not because of the consequences to myself, but because it didn’t end up working. The olive bar remains open so my act of defiance didn’t accomplish what I set out to do. If it had resulted in the olive bar being closed, I think I would have done it despite the consequences. Maybe there was something else I could have done to force the issue more, maybe I should have tried to outright sabotage the cooling mechanisms, maybe I should have called the local news, maybe I should have tried to convince my coworkers to go along with it to make it harder for them to get rid of the problem person, I really don’t know and hindsight is 2020. It’s always easier to tell after it’s too late to matter.