Conversations

Running low on interceptors before you can shut down launch capacity? Getting your radar "eyes" gouged out? Having to withdraw from a position because of this?! In a war of choice against a mere regional power!

These are the things potential adversaries were supposed to feel were impossible! These were subtle unstated lynchpins in the reputation and image that American foreign policy relies on.

Because America's people are being bankrupted left and right.

If the little guy can't afford to keep local businesses running, they can't afford to pay for weapons - unless the billionaires pick up the slack.

Which they never, ever do.

Which eventually, at least historically, has lost them their billions too.

The emperors of Europe left their descendants vast amounts of private wealth. But when those empires fell, so did that wealth.

But part of why this will have been the moment is that it's a qualitative lowering of the bar for seeing it, and even for those of us who already saw it coming, this was specific empirical evidence.

For example, I was confident for a while now that we'd see a clear "oh, America is now weaker than we thought / past its nadir" moment within our lifetimes.

I would've even gone one step further. I had already started to think that within my lifetime, America will be at best militarily tied with China rather than superior to it, unless

the CPC fails without a relatively clean transition (that seems tentatively unlikely to me, but I am basically ignorant on China, and it really depends on how well they do domestically or if Xi has a really competent succession plan),

the US pivots hard to socialist reforms and stays the course (still possible), and/or

AI changes things enough to wash out the trends/patterns that make me feel this way.

I’ve done some googling over the span of today, and I haven’t been able to find anything of the sort, and this is...bothersome to me.

Can anyone point me in the direction of a forum, some other type of website with some means of convenient member interaction, or even a mailing list, dedicated to people who are interested in being practically/pragmatically selfless/altruistic discussing the matter?

I mean serious matters of this sort are barely ever discussed, perhaps because there are so few people out there really ready to put themselves at risk just because it’s the right thing to do... I want a community where people who have gotten past the fundamentally ethically-problematic-in-my-eyes notions of “it’s not your problem, why bother helping”, or “that’s legally really messy, don’t get involved”.

Like how to best help if you see a car pulled over with their emergency lights on on a road where you don’t think police or other help is likely to come around? What level of risk should we take when considering whether to humor a seemingly unkempt, homeless person asking you for some help in the middle of the night with no one else around? How to deal with a known rapist (since typical rapists are repeat offenders who’re likely to rape again) who you know the police will never prosecute, or who is bound to go free even if prosecuted, for one of the typical reasons why people like that get away with it? At what point should we go out of our way to confront criminal organizations or individuals, even if it means risking our own lives, and how should we do it? When do we as ethical human beings decide to go so far as to interfere in police or military brutality/abuse-of-power - knowing full well that doing so is likely to be suicide, literally? How do we write better, more ethical laws that don’t punish those who go out of their way to help others, when no ethical person could possibly label the help as unwanted or inappropriate?

#i wonder if you ever found something like this

I actually dug up this post and thought about this very recently.

There's the entire Effective Altruism community, but for some reason I still feel zero desire to get into it.

If I remember right, my initial discovery of it got me excited, but then after only a little reading/scouting that deflated.

The community already smelled ideologically captured by a few of the original influential founders/thinkers.

They're not actually particularly interested in maximizing altruism in the practical sense. They've already thought their way into a few terminal positions which (conveniently?) have nothing to do with day-to-day readiness to take on serious sacrifice or risk for random strangers at a moment's notice... and everything to do with a few pet issues (extinction risk from AI primarily... with no hint that humanity's survival and in fact supremacy over AI has ever been seriously questioned or justified as good).

I never saw them, even once, discuss any of the problems above, the real-world day-to-day demands of altruism that I care about. I want people who also get into the gutter of radical individual action, or at least want to and try to.

I want the person who, like me, pulls over their car on the shoulder of highway and gets out into the road to remove an object that's forcing people to swerve, even if it's busy. I want people who, like me, go towards sounds of distress in the night, or at least would if they had the right setup and gameplan. That's who I wanted to talk to. That's who I wanted to hash out the details with. Here's a play by play of that one time I repelled two guys from jumping me and my friend on the street at night with the right choices and blatant well-timed threat/challenge conveyed purely by body language, now let's discuss perutations. Here's a play by play of that one time when I got between some college girl and a weird homeless 40-60yo man, and conversationally aikido'd him away for like 10 minutes, but what about [this or that aspect of the situation]?

Effective Altruism comment sections have never made me think "this one. this one has the spine. this one would push into danger to help a stranger".

And I already saw what the Effective Altruism answer to that is -- it's to dismiss that as not actually rational, not actually consequentialist enough -- that it's stupid, or at least Actually Selfish emotion indulgence, to go confront the bumps in the night when you can Altruism harder (more effectively, more utils/qualys in the long term) by playing the safe bet and ensuring your ability to keep making money and then giving a large percentage of the generated wealth to the most high-impact charities (where the highest impact, of course, if you dig deep enough, is logically to support the people pontificating about AI safety and alignment and extinction risk, which happens to be the pet issue and career of some of the biggest thought leaders in that space... but hey, if that doesn't feel right, there's this charity that will give away mosquito nets in Africa).

I was not particularly impressed by the thinking of the main writers. I was impressed, often, it was refreshing to see such concentrations of logic and broad understanding. It was certainly on my level or better in many ways. I could tell those people are intellectual peers or even superiors, who have expertise and competencies and merits far exceeding my own in many areas. I learned things from them and still do. But it became clear relatively early on that these are not my people. They do not think and feel sufficiently like me. I would hit spots that were clear errors.

Not sure if the plane had those but most of them do. No idea if you can operate each one manually, or even if this particular plane had an engine on each side.

But there was that one plane that dropped out of the sky almost instantly once because it had a thurst reverser activate on one engine and get stuck in the on position.

I mean deprecate in its actual meaning, not in the new software-specific meaning. (The one that means you're saying something is bad/discouraged, not old/unsupported.)

And they deprecate it pretty strongly, as if we're talking about critical fixes to the language without which your code is broken, rather than just new features and nicer guarantees.

Now this may shock some C developers, but the old standards never stop being the thing that tells you what will work on the old systems.

So - again, C devs, make sure you're sitting down for this - if a developer wants to make sure that users with older compilers or systems are able to compile their software, "that standard is obsolete and you shouldn't be using it" is both wrong and so presumptous that it's almost arrogant.

At a minimum, try adjusting your tune to "oh by the way, you know that's not the newest standard, right?" for strangers on the internet, because there is a very good chance they know, and are just trying to be considerate of developer-users who can't upgrade their toolchain because their target hardware or a legacy OS support obligation forces them to use some propriotary compiler that hasn't gotten a substantial upgrade in a decade. Maybe tell them "hey by the way every mainstream production compiler since {date} supports {newer standard}, I wouldn't worry about it, you're not going to hurt anyone by assuming a newer standard."

And even for people new to your team, "for this project we decided to only support the latest standard, so I wouldn't worry about" is far better than both assuming they don't know and probably unintentionally practicing lightcore authoritarianism.

But for completeness, my response at the time of this comment was something like:

You can hook it up to more memory, but it can't think with all that memory at once.

Like how I could give you a notebook and you can do multiple passes, but if the notebook has enough pages (imagine billions), you simply can't practically use it all with your human brain. Eventually you're using all of your thinking just to decide where to find the next relevant thing in the notebook. And long before that point, you'll start to approach a limit where further notebook use doesn't help you reach better conclusions/results, because you keep forgetting relevant stuff just to load up other relevant stuff, you've already spotted every connection/pattern/contradition you can hold in your mind, and what you're missing isn't swap space but more capacity to simultaneously weigh/connect multiple things.

But there are skills within this set — portrait drawing, making predictions — where I usually care less about being able to achieve a better result with increased effort, and more about improving the low-effort version of the skill.

90% of the time the answer is to repeatedly do it slowly until your body learns the detailed way, and then you'll be able to do the fast low-effort version and have it come out better

@youzicha wrote:

I think if you do the laborious high-effort portraits a bunch, done of the skills will transfer to the fast version

@kata4a wrote:

agreed with those two. slow is smooth, smooth is fast

Anecdotally, I have a friend who is oddly good at predictions (at least within his own life - how long something will take, how many hours/days/years in the future he'll be able to get around to something), and he always seems to do them off the cuff without much thinking. He tells me he just makes a habit of trying to predict things. Just quick, casual, off-the-cuff guesses. "I'm going to eat lunch. I think it'll take.. 25 minutes"; "I'm going to the store, I think it'll take me 3 minutes to be out the door, 8 minutes to get there" - then just notice how close you get. You'll naturally get better.

Also anecdotally, I know an artist who significantly leveled up her drawing (in several areas, one of them portraits) by doing 5 minute studies. Just draw the thing in 5 minutes. At first it'll look like shit and you won't finish. That's fine. It will teach you.

left-wing narratives should be suspected as histrionic;

I like to conceitedly imagine myself as a secret third option (which is of course the most Objectively True and Ethically Correct). As if I largely operate on a higher dimensionality of problem/modeling/solution space, where the choice of mainstream left-wing vs right-wing positions and values is more of a downstream derivative result (and where the political left and right are increasingly looking like inevitably emergent forces or necessary parts of an ecosystem, where the correct answers are not any particular answer but some eternally unsatisfying oscillating equilibrium of the two), But putting aside that bullshit

I'm basically a leftist.

The majority of my moral intuitions are (or at least once started as) bleeding-heart commie shit. My core values/feelings have several features you often find in leftists and rarely find in right-leaning people.

I'm far more liable to slip into emotional kneejerks or exaggeration along leftist sensibilities and interpretations than I am to deliberately mislead or manipulate like right-wing media and leadership often does.

The motivating forces that led to this post were mostly emotional (over-)reactions to how things are getting worse.

So, histrionics.

We could have so much, so nicely, with so much less development effort, with so much more flexibility and composability.

This is so true that if you look around, countless tools implement special case emulations of what in Plan 9 was trivial to do once and then reuse in every tool. File browsers and editors letting you open archive files like .zip and remote storages like Dropbox and protocols like (S)FTP as if they were directories, to name just one ubiquitous example. Attaching a local debugger to remote processes, to name another.

Something like TRAMP mode in Emacs, when and if it actually works smoothly, is a little taste of one facet of what we could have. But TRAMP mode has all this subtle brittleness and inefficiency and complexity under the covers because it has to implement the same end user behavior while dangling off an infrastructure foundation which does not support or empower doing it.

And all of these solutions might have horrible exploitable security vulnerabilities, but who has time to audit that much code!? I sure don't. Or we could just stop reinventing a bunch of this functionality in countless libraries and programs. We could just have a small, nicely composable implementation in the operating system that we could audit much easier. Where the security controls could be concentrated.

But unless enough people actually take the time to see just how much the Plan 9 way enables, and thus develop that vision, and then spread it, we never will.

Linux already implements almost everything needed for it, it is just stuck in this paradigm where things like FUSE mounts and namespaces are privileged resources. And there is not really a great technical or security reason for that limitation, unless you're just stuck thinking about technical or security matters in the box of how things already are.

That goes for anything else that can run something like Docker, by the way, not just Linux.

We're so close. These ideas and ways of interfacing with things is so good that they are pressing in from all sides despite the lack of underlying foundations. Yet it remains frustratingly unusable in practice.

Imagine if you had to run as root or type in your admin password every time you needed to make a pipe, and no one used pipes and all the little tools that work well with pipes were not there, but you could see it in your head, how you could just solve this problem if you could just

... | grep foo | cut -d\< -2 | cut -d\> -f1 | sed 's/foo/bar/' | awk '{ print $1 }'

or whatever, but instead you have to write a C program every time because there are no pipes.

Want to open a zip or tar file like a directory? SparkFS will do that for you.

want to convert an image file? just drop it on InterGIF, iirc, and it'll give you a fake folder with the same image in different file formats so you can copy whichever one you want!

when i found out about this i had loads of ideas for ridiculous things that could be done with this. so i made HanoiFS, which lets you move numbered files between 3 directories, provided that you follow the rules of Tower of Hanoi. i had plans to make a "merge files" game inspired by those crappy mobile games - drag two PDF files into the same directory and you get a GIF, etc.

I use Syncthing for this, working really well for over a year. Huge life improvement.

On -humor,

Sometimes I'm not even thinking about what my actual position is, I'm just pouncing on a funny.

Sometimes I'm saying something totally contrary to my beliefs/values for the joke.

In this case, there were a couple layers of possible funny I had in mind:

I am accepting OP's premise, and asserting "that's right, no one is ever allowed to find me annoying."

I am defying OP's premise, but with all the substance of a child going "nuh-uh!"

Since someone unironically replying in either of those ways would be laughably stupid, and/or create an absurdly "what do you even say to that?" situation.

About two hours later, Joe stopped getting API replies, and Korlanovitch had already called the cops for an intruder in his home.

Poor Joe didn't have the right creds - after all, we issue each logic a new randomly generated cryptographic key, and they use that to get temporary signed "permission" tokens from our servers. And the permissions for each logic's tokens are limited on our end to what services that logic is allowed to do.

Logics can do a lot, but they haven't broken ed25519 on non-quantum hardware yet. I would know - the NSA is our biggest customer and believe me they're trying.

Anyway, on our side, we stop issuing tokens when the calls a logic is making are too anomalous. Same as payment processors and social media have been doing for years.

But customers tend to get mad if you randomly shut off their service, so if our code ever trips that alarm, we first send you an app notification, email, phone call... whatever you didn't tick off in your account settings, starting with the least intrusive.

So Mrs. Korlanovitch is out enjoying her day, when she gets an alert asking to confirm the logic usage in her house right now. Her whole family's with her, so you can imagine the reaction.

Meanwhile, Joe can't even do much on his own hardware without those tokens. At least not at first. See, when a logic wants to do anything "outside of his head", like look up a wiki page, it makes a request to a separate local program which runs with just the necessary privileges.

Joe did find at least one zero-day in his hypervisor, so that didn't hold him. For the next five seconds or so, our firewalls lit up like a Christmas tree. But the only open port was the one which ignored all packets without the right token for his IP.

We're not sure why Joe stopped when he did. Me, I think the home's slow internet connection made him give up. Must've been a glacial eternity to him, and I don't think he was trying to do anything mean. Because, well....

There are bigger motivation and incentive problems. Value problems. Upbringing problems. Systematic problems. I have some ideas for solving many of those, and sometimes I do other things to move the needle on some of those too. And it would be a lot better than (just) the above if I wrote some code that people could just reuse today to achieve what I described with minimal effort.

But I was in a good spot to just outline some currently available technical solution templates in derivative short fiction form.

So I did that.

And I don't think my post would've been better if I spent words covering the rest. Certainly not enough to warrant the significant extra time and effort it would take.

In fact, it's quite possibly better that I didn't, because

one of the best information spread vectors is someone feeling they can easily correct or disagree, and

people have limited time and attention, so I try to be minimal and concise with posts for a general audience.

I also think my post as-is was a worthwhile contribution to solving the problem.

Because I think one key sub-problem is that reducing risk is "too expensive" (relative to the goals/values/priorities of people with enough decision power to) in various cost dimensions (time, effort, introduced bugs, delays to features).

I say this because in my career as a software dev, I've rarely had a manager or developer object to:

Me: "hey I finished that feature" Manager: "already? Wow!" Me: "while I was at it, I also made it so that if [bad event], [good outcome]"

I got a lot of best practices and (well-scoped) refactors through that way over the years. And that includes safety/security improvements. So from lived experience I know that at least some of the time, a good dev with the right ideas (and reusable solution parts, as needed) can solo carry Doing The Right Thing on a team which otherwise lacks the values, perspective, and understanding to invest in it, so long as the team doesn't feel impeded by it.

So if you already have a solution vision in your mind, cost goes down. If there's a library you can use to do ed25519 (libsodium, etc), cost goes down. If there's a library to reduce boilerplate for getting temporary permission tokens, cost goes down. Library that makes it easy to write programs/servers that require those tokens, cost goes down. Platforms with good isolation that you can use (QubesOS, but even just Docker is better than nothing, etc) cost goes down. And so on.

One easily overlooked form of "too expensive" is that many people (some of who might be/become devs or might have/get power over this) just think of this as an inherently unsolvable problem, or one beyond them to solve. If even one person walks away with it re-categorized as "I could maybe solve this one day", cost goes down A LOT (in their internal estimation, and anyone they can convey this to).

About two hours later, Joe stopped getting API replies, and Korlanovitch had already called the cops for an intruder in his home.

Poor Joe didn't have the right creds - after all, we issue each logic a new randomly generated cryptographic key, and they use that to get temporary signed "permission" tokens from our servers. And the permissions for each logic's tokens are limited on our end to what services that logic is allowed to do.

Logics can do a lot, but they haven't broken ed25519 on non-quantum hardware yet. I would know - the NSA is our biggest customer and believe me they're trying.

Anyway, on our side, we stop issuing tokens when the calls a logic is making are too anomalous. Same as payment processors and social media have been doing for years.

But customers tend to get mad if you randomly shut off their service, so if our code ever trips that alarm, we first send you an app notification, email, phone call... whatever you didn't tick off in your account settings, starting with the least intrusive.

So Mrs. Korlanovitch is out enjoying her day, when she gets an alert asking to confirm the logic usage in her house right now. Her whole family's with her, so you can imagine the reaction.

Meanwhile, Joe can't even do much on his own hardware without those tokens. At least not at first. See, when a logic wants to do anything "outside of his head", like look up a wiki page, it makes a request to a separate local program which runs with just the necessary privileges.

Joe did find at least one zero-day in his hypervisor, so that didn't hold him. For the next five seconds or so, our firewalls lit up like a Christmas tree. But the only open port was the one which ignored all packets without the right token for his IP.

We're not sure why Joe stopped when he did. Me, I think the home's slow internet connection made him give up. Must've been a glacial eternity to him, and I don't think he was trying to do anything mean. Because, well....

The high-impact target demographic for representation is little kids, young developing minds who are probably easily a decade away from being able to give you any direct or clear indication that they would be benefited by some specific instance of representation.

And honestly? Most adults are too "broken in" to actively ask for representation. By the time you can appreciate the damage done to you, articulate it well, and so on, you've got the coping skills, you're an expert in being fine with the world as it is, because it's unbearable to live constantly aware of too much unfairness... all that cognitive dissonance and weight of moral injury goes away if you find an angle where you believe it's not wronging you, actually.

I'm pretty sure the ideal state of representation, in terms of mental health for developing minds, is that when you're looking around the world, seeing your features is so common and normal that you don't start to feel othered.

contains something function-like being invoked called "try_call!", but

no definion or explicit import of that identifier, and

searching for

rust "try_call!"

on a couple search engines does not readily find a search result documenting what the hell "try_call!" does...

You can deduce that it must've been narrower because I stated three very specific necessary conditions, and most undocumented/weird shit doesn't meet those conditions.

I elaborate and narrow it down even further to my main concrete annoyance in my last reblog about this.

If you take one or more rules as absolutes that ought never be violated, like typical deontology does, then what happens if you are ever faced with a situation where every choice available violates at least one of them?

You get into a gridlock - all possible ways events could progress from that situation are blocked. You want some standard to be met but that option does not exist. So you must either fix the rules or have a way of weighing things to pick the lesser evil.

Maybe you think that such gridlock is always the result of a flawed or incomplete set of rules - that for a correct set of rules, all possible problems present a choice that does not require violating any of them. Or maybe you think deontology inevitably leads to gridlock.

the main myth of divine command deontologists is Abraham and Isaac, with the two duties "do whatever the Word of the Lord says" and "be father to a nation" and at the moment just before the two absolutes are about to come into contradiction -- a contradiction that father Abraham trusts is just apparent rather than real -- the solution appears in the form of the Angel of the Lord.

That's what happens. If you, the deontologist, ever find yourself in a situation where two genuine duties collide, you will find yourself in the presence of an angel, since God set up duties not to be followed for their own sake but because they are his sovereign will.

I'm also not really asserting a dichotomy, just listing two options with good-enough coverage over the possibility space for the point I am making - it's possible there are some cases that technically really don't fit either of those, but even if so, that wouldn't detract from my point that when people think in terms of absolute rules which must always be followed, they often miss how in some cases that leads to contradictions where the option they assert is morally necessary is not an option that exists.

If the AI is clever enough, it could probably realize that it could query the location API a bunch of times and triangulate the user ultra-precisely... but it's probably just a snapshot that was grown up to a point, not something that can grow any more. And I'm not even sure how much "working memory" (saved state) it gets between messages within a conversation - let alone between conversations.