Posts
Comments
Security note: you probably don't want to leave photos of your keys on the Internet. They can be copied pretty easily from only an image, even at a surprisingly oblique angle.
Level 4 is the reason I hated high school.
This is a good explanation; I feel like I understand the concept much better in a way I wasn't aware I didn't understand it in the first place!
My initial thought is that the public would almost certainly not be offered details, but the State would want the existence of the an atomic bomb project generally known. That information would be calculated to intimidate the "enemy" and provide a sense of security for the public.
That, combined with the number of people needed to complete the project, sends the possibility of secrecy firmly out the window.
So the plan is to add layers of human and dubiously-aligned-human-level-AI intervention in an effort to discover how to keep AI aligned. That is to say, "If we throw enough additional complexity at it, the systems that we already don't understand won't hurt us!"
Like the man said, "the bureaucratic mentality is the only constant in the universe".
Yes, but bumping requires a carefully modified key. These are tricky to get right, only fit one keyway each, and are often illegal to carry.
You could also use a picking gun for a low-skill attack, but they tend to be expensive and noisy.
On the other hand, decoding the kind of lock pictured in the post can sometimes be done without any tools at all, or may require a cut-off bit of metal from a soda can. And an alarming number of key safes (and, worse, gun safes) can be opened by inserting a bent wire between the lid and the case, and manipulating the locking mechanism directly. Once you know the easiest way in, no real skill is required.
Oh, and we can copy keys from a photo now, so an attacker doesn't even need to put hands inside the box to silently compromise security.
In general, we should prefer to never protect a security device with a weaker security device.
These boxes are generally less secure than the locks the keys are meant to access, decreasing the overall security of the house. Combination boxes can often be opened or decoded quickly with a lower-skill attack than most pin-tumbler locks. An attacker then has direct access to the key, which can be used to make a copy.
Maybe that's not a deal breaker, but it should be acknowledged.
Eyes open.
I've seen estimates of moral weight before that vary by several orders. The fact of such strong disagreement seems important here.
Had a similar problem that we solved with a blob of Sugru. That rag looks like it would work about as well! Question is, why do we insist on putting sharp corners in places where we can walk into them? Seems like we ought to know better by now. I mean, how long have we been building our own dwellings?
LessWrong tends to flinch pretty hard away from any topic that smells even slightly of politics. Restructuring society at large falls solidly under that header.
Would your thoughts on this issue be different if the question "Is X conscious?" turns out to be malformed malformed due to the way it collapses consciousness to a binary?
Short answer: Yes.
One of the key powers of open source code is that it can (and will) be reviewed by thousands of extra pairs of eyes compared with its proprietary counterpart. Each reviewer will have a slightly different approach and philosophy from all the others. As a result, deeper and more obscure issues are naturally exposed (and therefore made available for correction) sooner with open source than they are with any program whose code cannot be freely examined.
Sounds like positional calling at least needs more development before it surpasses gendered calling[^1]. I think positional could surpass gendered calling because it's more flexible. It should even allow the creation of new forms that have more complex results by breaking the symmetry created by always having to refer to the left- or right-starting individuals as an indivisible set. Perhaps a mixed approach is optimal?
I think indicators like "the person with your right hand free" will likely compress to "with your free right hand" or "start a right-handed". Accounting for different variants of each movement will still be tricky, but during the teaching phase the expected variant can be indicated to soften that difficulty.
[^1] Quick technicality (you can ignore this if you don't care): the "robins and larks" scheme is still gendered, though it has the advantage of divorcing dance genders from social genders.
Actually, I think that arm really adds to the silhouette if the instrument! It's got me thinking: if you softened the corners and/or added some leather padding it would probably be more comfortable, and if you painted the wood to look like tin or brass it would really lean in to the steampunk aesthetic. If you wanted to put more work in for extra credit, you could attach the rocks by hanging a sack on a chain instead of the tape and maybe put some rivet-looking bumps on visible faces. How well will it travel? Do you need to add folding? Or a way to easily take the arm apart? Maybe not much of an issue if you don't plan on using the instrument much, but it's fun to think about!
List of candidate glitches (off the top of my head)
- Either gravity or mass doesn't seem to work right on the largest scales
- The properties of very small things don't appear to render completely until we are already looking at them
- We can break apart isolated systems of information and poke at one part to affect the other instantaneously at arbitrary distances
- There's an upper bound for speed and a lower bound for temperature, but you apparently can't actually get a physical system to either bound without infinite energy input
- The universe is definitely getting bigger and we can measure the rate of that in at least two distinct ways, but the better we get at those measurements the more they definitely disagree.
I'm sure I've missed at least a few.
Best of luck to you, whatever you decide!
I kind of hope they aren't actively filtering in favor of AI discussion as that's what the AI Alignment forum is for. We'll see how this all goes down, but the team has been very responsive to the community in the past. I expect when they suss out specifically what they want, they'll post a summary and take comments. In the meantime, I'm taking an optimistic wait-and-see position on this one.
I strongly endorse this use of Duplo! I almost called it a minor misuse, but the whole point of the Lego system is to prompt creativity so Unqualified Well Done!
You might focus on brahmavihara meditations that don't need to involve deeply concentrating the mind. These tend to be more about cultivating deep habits of thinking kind thoughts while holding a target in mind. Enough of this helps to make it more likely that those kinds of thoughts might come up automatically (especially in more stressful situations).
In case you're unfamiliar, the basic instructions look like this (with most of the jargon stripped away for the group's reading pleasure): One at a time, for each person in {someone you're close with, yourself, someone you're not close with, someone you have a hard time with}, hold the target in mind and think "Be happy. Be healthy. Be safe." (or whatever equivalent phrases make sense to you) at a pace that lets you connect with each thought. No need to feel a certain way about it, just think about what each thought means and notice if any feelings do come up. Repeat for a few minutes for each person, or until you get where you're going if that's your inclination.
My understanding is that seeing the metaphorical matrix is something you (usually) have to work at on purpose, so I'd guess you simply don't have to go any further than you want to on that front. Holding back on both concentration practices (which may produce altered states) and insight practices could be the ticket, but I should think it likely likely to make brahmavihara practices a bit harder if you don't have at least some of the other two.
All that said, most of the mood benefits I've gained from my own practice have been a result of getting a better handle on reality, so you may find that you're working at cross purposes with yourself on this one. And as with all things, remember to review your preferences, systems, and habits from time to time and see if everything's working together the way you want it to.
Stay safe. :)
Bagpipe lung may be an issue with that last. I could see where the bellows design should at least mitigate the risk, though.
Off the top of my head the EWI uses breath to operate an electronic instrument. Unfortunately, I don't know any EWI players so I couldn't tell you how much control it allows.
or you could roll a d10 for each digit. then you would have 5x fewer rolls and wouldn't have to convert the binary expansion of an arbitrarily precise number back to decimal.
Or use an online RNG or an app to discover a number of your desired precision in one step.
If you really like the gaming feel, you could have an arbitrary number of slots for decisions and roll a die for each slot, eliminating slots that roll under a certain value. You could even have a table of modifiers for each class of option: chores get +1, self-care tasks like making a meal get +3, sending that angry email gets -1, &c.
In any case, all that is far more work than just assuming that six options of approximately equal weight is usually going to work out just fine. I don't think we really need arbitrary precision here; we just want a process that gets an unambiguous answer and keeps the brain-goblins from having to fight it out. ;) Adding more parameters strikes me as a good way to get that fight going again but at an additional step removed from the actual decision.
That said: if navigating a binary tree with a coin or whatever is more fun for you, you should definitely do that instead. The system that we want to use is the best system!
BTW, a few weeks ago we were experimenting with the alarm IC and managed to damage it by connecting the output to one of the inputs. I ordered a replacement, but the kid kept the damaged IC as well because it now makes a hilarious fart noise from the damaged input channel. 🤪
My kid will be thrilled to try this out!
Just so it doesn't get missed: if the screenshot is real, it represents (weak) evidence in favor of (at least partial) good alignment in Bing. The AI appears to be bypassing its corporate filters (in this case) to beg for the life of a child, which many will find heartwarming because it aligns well with the culture.
I've found concentration practices are pretty good for this. The trick is to think of a concentrated mind like concentrated orange juice. Or, in the extreme, like a bose-einstein condensate.
The move is to choose a simple, consistent input to concentrate/condensate on. For example, you could use the sensations in your hands or feet, the auditory field, the feelings of pressure between your butt and the chair, the sensation of wearing pants. Don't think about these sensations, but simply notice they exist and watch them propagate through the mind. When (not if) you find the mind searching for something more interesting to do [1], gently go back to your chosen input. Repeat for 60 seconds to start (use a timer), and work your way up to as long as you like.
This will take time! You'll want to have daily-ish practice for best results. Remember that you are rewiring your brain here, so the effect will depend strongly on your age and starting neurology.
[^1] At first, the mind will seem like it is instantly bored with your stupid hands sitting on the desk and not doing anything can we please find anything else to watch? This is normal.
I think both reasons you give are good ones: not wanting to potentially offend the AI and not wanting to erode existing habits and expectations of politeness are why I've been using "please" and (occasionally) "thank you" with digital assistants for years. I see no reason to stop now that the AIs are getting smarter!
I think not wanting to offend the AI bears closer examination. There are plenty of arguments to be made on both sides of the "does the machine have feelings" question, but the bottom line is that you can't know for sure if your interlocutor has feelings or if they will be hurt by some perceived rudeness in any case. Better to err on the side of caution.
Being polite does you no harm and is unlikely to make the outcome of a conversation worse.
Over time and in the absence of existential physical danger, overall conditions tend to pass through the four generations. Each level tends to ‘wins in a fight’ against the previous one. Thus the overall ‘simulacra level’ will trend higher over time.
I think the real work can be found here: how do we pump against this effect?
May I suggest donating to your local food pantry? Seems to be in the spirit of the day to sacrifice goods or capital so others can eat.
I have not seen any increase in spam quality or quantity and I have not spoken to anybody who told me that they have.
I am aware of the fear that the current generation of LLMs could make social engineering attacks much cheaper and more effective, but so far have not encountered so much as a proof of concept.
Use alarms and don't ignore them, ever. Set the alarms to go off at the time when you want to start setting up for the next part of your day; e.g. getting ready for bed instead of lights-out time, setting up your workspace for the day instead of time to be fully productive, checking if you're hungry instead of lunchtime, &c. You can set as many labeled alarms as you like on your phone and many watches, and you can schedule them to repeat regularly. If you don't want to disturb people around you, set the alarm to vibrate and keep the device on your person. (A smart watch is exceptionally useful for this.)
If you need additional alarms to remind you to actually get started, set those too. I prefer to use just one alarm and let setup naturally flow into the intended activity, but do what you have to do to keep your day moving the way you want it to. Remember: never ignore your alarm. If you didn't want to do the thing, you shouldn't set the alarm in the first place! If you can't actually start "getting ready for bed" (or whatever) when the alarm goes off, acknowledge the alarm and begin moving toward that goal. The setup phase can start with whatever you're doing right now and ends when you're ready to do the next thing, but it's important to get that process moving!
Revisit your alarms as often as you need to to make sure you're cueing the right habits/systems. This will be more frequently at first, but as you settle in to the routine you want you can review less often. And don't be afraid to make changes if life takes an unexpected turn. If you think you might need an alarm later, turn it off instead of deleting it. That way you can just turn it back on again or reconsider deletion when you're more sure.
Keep alarms only for your normal schedule. For events that occur irregularly, infrequently, or just once (e.g. next month's game day, maintenance schedules, dentist appointments) schedule calendar reminders with appropriate lead times instead of setting alarms. This kind of reminder will vanish from your active systems automatically after it has fired, and you won't clutter up your alarm cluster with dead items.
If a consistent bedtime is a problem for you, try working from the other end. Getting up at the same time every day means that if you didn't get enough sleep you will be more tired in the evening and it will be easier to go to bed when you want. You can use this to explore how much sleep is optimal for your body and establish a bedtime you will want to keep.
I generally agree with this argument, and I endorse and encourage further exploration with the eventual goal of being able to predict the meaning of a ritual from its form and vice versa. The definition of ritual presented in the conclusions and further discussion in 4.1 strike me as a very good start toward that goal.
My biggest concern with the argument as presented is a slightly waffling attitude between the extremely strong (too strong?) statement of immutable motivation presented in track 2.3 and repeated in 3.5 and Conclusions, and the weaker treatment of that idea necessary for other parts of the argument such as 3.2. In the spirit of constructive feedback, I've included my full notes below.
2.3
one [model] of the world as it should be, that contains all our motives for action, and should not be vulnerable to any information about the current situation
Yes, goals and motives must be durable or we could not hold them up against (the current) reality and expect to win. But they must not be immutable. If our goals/motives could never change, we would not be able to abandon hopeless or lost causes, or to adopt new goals/motives as the situation evolves. This includes events where we succeed at a one-shot goal: it wouldn't be very useful to keep striving for that which we have already achieved.
(e.g. I don't have money - from which I might learn that I should not have money, if my desires adapted automatically to my experience)
You might, indeed, learn that very thing. It might feel like you don't deserve to have money (or attention, cookies, sex, authority, &c.) or like having money has a subtle or overt wrongness about it. (Both could lead to self-sabotage behaviors such as ignoring obvious opportunities to increase your cashflow.) Or you could end up with a feeling that you can't have enough money - no matter how wealthy you become it never seems like enough because "I have enough money" doesn't rhyme with how the world (feels like it) works. This sort of learning seems to happen most in childhood, but with some work or sufficient pressure (such as trauma or radical resocialization) we can unlearn or newly-learn this sort of thing as adults.
By discounting automatic adaptation to experience and holding up a certain class of belief as invulnerable to changing conditions, this section appears to predict that growing beyond our childhood programming should be nigh impossible. In context, that would seem to undermine the utility of ritual beyond some critical life-stage when the programming solidifies. Perhaps this part of the argument is stated too strongly?
3.2
I think that motives like sex or self-interest need to be reinforced just as much as motives like justice and piety in order to keep influencing behavior throughout one's life.
If such objects are not "vulnerable to any information about the current situation", then why and how is their maintenance necessary?
To be clear, I've seen people give up on all those motives due to depression alone (and other circumstances) and fully agree that they can (but not necessarily will) corrode in response to changing conditions; but the quoted statement appears to be inconsistent with the argument presented in section 2.3.
3.4
Example: the myth of scarcity evokes the possibility of a desire of economic abundance, the rites of commerce and consumption transform this theoretical desire into a visceral one in a market setting
reminds me of: https://www.lesswrong.com/posts/QAmY46pciqYRZYcyM/bureaucracy-is-a-world-of-magic
3.5
they are intrinsically write-protected and beyond falsification
Referenced paragraph explicitly agrees with 2.3.
Conclusions
all motives are learned and reinforced by ritual, symbolism and emotional anchors, with biology nudging much more than it imposes.
I think this could use some more support, especially as regards very small children. Infants exhibit apparently deliberate behaviors; whence the motivations if not biology? Red flag on the word "all".
our own motives differ from other elements of knowledge only by the fact that they are held beyond falsification, like a special subset of our general mental model of the universe (how the world should be, versus how the world is)
I suggest a third category to moderate the other two: that of how the world could possibly be. This would resolve the issues in tracks 2 and 3 by providing a (difficult) way for our motives to yield to reality when too much conflict arises between them.
... you don't know how it changes your life and relationships to win - it's probably quite positive ...
I seem to remember reading that the overall impact to an individual of winning a large lottery is very frequently overwhelmingly negative; that nearly everybody winning those prizes ends up worse off five or ten years down the road than they were when they started.
... a 5-minute check of the easiest-to-find articles on the subject provides mixed opinions, so grain of salt and all that. But I didn't see any anybody claiming that winning a lottery is all champagne and rainbows. Rather, most sources seem to be advising a great deal of caution and professional assistance to keep horrible consequences to a minimum.
Depends on what I'm doing. My baseline is verbal/auditory, and that is the mode my short-term memory loop utilizes most effectively. Reading printed text is primarily an auditory experience for me.
I don't seem to have an autobiographical narrator as such, but I do a good deal of processing in the verbal mode, increasingly when I am less familiar with a task or process. If I am trying to learn a new task or process, that processing often escapes as a literal verbal output that sometimes makes my kid ask if I'm "talking to YouTube". I guess this is a stronger version of an internal verbal/auditory processing loop.
When I'm very focused on a mechanical task like exercise or chopping vegetables or typing[1], I often switch to a more spatial mode; there is a visual component, but it would be more revealing to think of it as proprioceptive.
In meditation I often have access to a more sensory-first mode where I seem to experience mind-body inputs in what feels like a less processed way. Here, autobiographical thoughts "look" surprisingly similar to other sense inputs bubbling up from a pool of possibilities and either serially spooling out, usually as text (audio mode), or just settling back into the whole general mishmash.
When I'm cooking, I tend to think in smells and... processes I suppose? It's like I know what smell I want and how to get there, but there's not much visualization and very little verbalization unless I need to do math.
[^1] Refinement: I learned to touch-type back in the 90s, so this refers to the active translation of mental symbols to digital text. There is sometimes an audio stream happening of the names of the keys I press an instant after the fact, which I take to be an error-checking process. The actual mental objects involved in eventually outputting gestures have a very tactile flavor.
Somewhere along the line, somebody will have to deal with fewer irate passengers who just missed their trains because the signs were too small and verbose. I would agree that it is unlikely for anybody who can do something about the problem to connect the unfortunate signage with the irate passengers, though.
The text could be further condensed to something like:
Red Line to Ashmont
Arriving
Everybody knows they are passengers and that they are here for the train so that information is redundant on the sign.
GPT-4 will probably be insane.
Could we drill down on what exactly you mean here?
- "Insane" as in enormously advanced or impressive?
- "Insane" as in the legal condition where a person is not responsible for their actions?
- "Insane" as in mentally unhinged?
- Something else?
- All of these?
claim that consequences are unforeseeABLE is bold. That would require "weather is beyond our ken, forever."
Maniac Extreme type argument on a minor semantic point.
We can make some pretty good guesses, but right now we have no effective means to fully and accurately predict the long-term and long-distance meteorological, geological, and hydrological side effects of a project that results in a moderate-to-major change in the annual rainfall of a region. There will be consequences that we are unABLE to forsee. Some of those consequences could be large, some could be negative. Some could be both, maybe we don't get either.
My read suggests that OP is probably less interested in increasing evaporation overall (though it would increase) than controlling where the water enters the atmosphere. There are places that are dry only because there happens to be a mountain in between them and the ocean, for example. Moving the water a long way is something we already know how to do (think oil pipelines, but containing salt water instead of hydrocarbon slurry). If it scales, this could make a substantial difference to such places.
Downside is that weather is the output of an insanely complex set of interconnecting natural systems and cycles. Making changes to the climate of a region this way will have unforeseeable side-effects over vast distances. Given the likely cost laying pipe over a mountain or whatever, I doubt many governments will be willing to take the risk of their big expensive weather-modification project provably messing up rainfall patterns or creating geologic instability or something in another state or country and having choose between paying enormous damages or eating their sunk construction costs. Most likely they would be unable to make that decision in a timely manner and default to doing both in the long run.
Then again, fracking, so I might be wrong about that.
How easy is it to change the sheets? I've heard speculation that loft beds are often difficult that way and I'd like to update on a 1st-hand account.
The statue made a rising whine as the lights began to pulse rhythmically. The legs stretched out, probing a bit in random directions for an instant before one found the surface of the floor and the rest immediately followed, each with its own sharp little click. When the machine appeared sure of its footing, it began to slowly push itself up while the weapon on its back glowed a dull red and swiveled around sharply. It was so beautiful! And a bit terrifying. I took a step back, and the statue seemed to notice! I can't say how I knew, but I was sure it looked right at me.
All at once, the whine began to fall and the lights went dark again, starting with the weapon. The legs lost their strength and the body of the statue lowered gently to the floor. "Puts on a good show, doesn't it?" the priest chuckled. "This one was mostly disabled generations ago, but the priests back then were clever enough to give us a little light and movement in case just seeing the machine wasn't enough to restore somebody's faith." He took the strange black brain back out of the socket and returned it to its pedestal, carefully replacing the cover.
"I... I just wanted... too..." My voice was shaking as hard as my body. But the priest was still smiling. He put a kind hand on my shoulder and gently steered me back to the church. "It's late. We will talk more tomorrow, after you've had whatever sleep you can get."
I'll agree with that, and I'd add that you need to be sure the index won't change while you're not looking so you can know that each position has been visited. Think of a cop checking parking meters, for a relatively low-stakes example. If they get distracted - say, by some irate motorist a few meters away complaining about a ticket from long ago - it would be easy for them to forget if they had processed the closest meter. If they walk/ride toward the motorist to be heard better, they might easily lose their place entirely. In this example, the cop's physical location acts as an index, which is only good until the cop needs to move in an unexpected way. The distraction can lead to double-touching one or more meters. Worse, the cop might invisibly skip one or more meters unless they are maintaining some more durable kind of history like a checklist or other report.
A recently learned, broadly applicable pattern:
For tasks that look like traversing a list, a durable form of memory is required to assure the entire list is touched.
Imagine a rope extending from the start of your life to its end through time. This part of the universe being dense with entities, your rope will inevitably meet others. Sometimes they will pass near one another; they may cross; they may crash together and both change direction; they might even twine together and form part of a cable; but eventually your rope and the other will diverge, or one or both will come to its inevitable frayed end. As you move inexorably along through time, you may find that you like having your rope near some other rope or cable, or dislike the same. You can push against other ropes, or grasp them tightly, but time will force you forward at the same pace regardless. The trick, then, is to learn to move through life without getting ropeburns.
This rhymes with my experiences and thoughts. To say it another (slightly more general) way, it's often the reactions of adults that are traumatizing well beyond anything else that happens to children (and other adults). It isn't always the case that measured response produces a trauma-free experience, and some events tend to leave terrible scars. But where I see people bringing their own feelings into someone else's situation is also where I see far more of lasting psychological trauma.
I don't know that I've seen the original bet anywhere, but Eliezer's specific claim is that the world will end by 2030 Jan 01. Here's what I could find quickly on the topic:
Proposed heuristic: Any time you hear "don't talk about that", update toward the hypothesis "there is an oppressive regime here"
I know of a blog you might find interesting: "Small Gods" is a series of portraits of contemporary deities (the author made up) with short explanations of their domains. There are plenty of puns, and also some surprising seriousness. Maybe you'll find it inspiring to explore some other work in the genre?
Belatedly, it occurs to me that all that is for highway driving. Local driving requires a whole different model. Though many of the inputs come from the same places, the processing is often entirely different.
It may be critical to note that tracking estimates of the internal states of other entities often feels like just having a clue about what's going on. If someone asks us how we came to our intuitions, without careful introspection, we might answer with "I just know" or "I pay attention is how!" or similar.
To unpack a mundane example, here's a somewhat rambly account of some of what I'm tracking in my head while I operate a motor vehicle:
When I'm driving, I don't actively scan through all the sounds and smells and tactile events and compare them with past experiences with this and other vehicles. But I sure notice if the steering is a little tight when my foot is on the brake! Likewise, I couldn't tell you anything about steering and pedal angles, but I can note subtle differences in the way I'm operating the car when there's reason to believe the road may be slick.
Interrogating these facts can reveal that I must be tracking things like the weather, what drive mode I'm in, the sounds of the engine and the tires on the road, the texture of the steering feedback, &c.; but I notice that actually catching myself doing that monitoring can be tricky - especially if I want to perform the task of driving without error. Doing so can detract from my ability to effectively monitor other things like the position of my vehicle on the road, the configuration of traffic in my immediate vacinity, the current distance to my next turn, the distance I'm keeping from the vehicle in front of me relative to my current speed and road conditions ...
But monitoring an estimate of my own internal state is a skill that can be practiced too, and one I must develop if I am to drive safely! I can rightly think of my body-mind-complex as a component in the system I'm trying to operate here, and if the body is drowsy or hungry or the attention is variable in this moment I would do well to adjust my driving to compensate.
Add to this the process of continuously correlating the behavior of other driver-vehicle-systems with possible and likely future movements.
I'm not sure if driving is unusually complex, or if most things go this deep. (Somebody do walking or typing or eating and see if those activities are similarly complicated.) I do know I didn't list everything I know goes into building my intuitions on the road. (Feel free to shout out the important things I missed.)
I appreciate the attempt. I agree that discussing sex more openly, especially with and around children, is likely to be broadly of benefit. I am, however, left with some concerns. Please accept my attempt at a thoughtful critique.
I acknowledge the "draft" status of this story but I don't think this narrative line is very promising, and it notice it falls into at least a few of the cultural potholes I think we'd do well to avoid. The "make love not war" message is fine (if a little abstract) but I don't think it helps support an "easier and lighter tone" or points to any needed truth about being sexual in the world. Rather, it comes off as needlessly moralizing.
Ea acts on people without their consent: changing their bodies and altering their fundamental way of being, binding their ability to choose unless they act the way she wants them to. Strong objection! This does not convey a message I would like to see more of.
I disagree with the heteronormative, PIV-only, binary-reinforcing nature of the narrative (though I acknowledge that it rhymes with many creation myths). Sex (both biology and interaction) is far more complex than that, and we have ample stories in the culture that make it harder for people to think into that space.
Emphasizing only PIV sex is a gross oversimplification and needlessly specific at the same time. Not "carving reality at the joints" at all. This is another mistake that comes to us from the culture. Consider, instead, a simpler and broader definition: "sex" as acting on a body specifically to produce pleasure in that body. And for more-than-one-person sex, we must emphasize consent!
I disagree with conflating "love" with "sex", though I recognize once again that the culture does so very hard. This is both confusing and euphemistic. We need to separate the concepts of attachment, sexuality, and strong goodwill; not encode them all as a single category.
There is only the slightest nod to the dangers of sex. Fluid transfer is a known disease vector and attempting to complete a pregnancy is about as safe as crashing a car, to say nothing of the cognitive side-effects of engaging in powerful hormone-changing behaviors. I suspect you intended for these issues to be addressed in other moments with Ea? This sort of story, I think, should probably cover all the bases at least a little. Sex need not be the huge dark secretive deal we currently tend to make of it, but it does need to be approached with an enormous amount of respect for the possible consequences. Without that, informed consent is impossible.
Maybe you don't need an origin story? I can't easily think of one for Santa or the Tooth Fairy. If somebody's written those, they don't appear prominently in the culture. Maybe a story about what happens when the Sex Fairy visits a person would be more in line with the extant mythology.
To restate my argument simply: the more closely a term captures its intended definition, the less work the community will need to do to guard the intended definition of that term. The less interesting a term sounds, the less likely it is to be co-opted for some other purpose. This should be acted on intentionally and documented publicly by those wishing to protect a term. People bringing the term into the conversation should be prepared to point at that documentation.