geoelectric

I just got an Ioniq 5 a little over a month ago and did pay with cash (ie no tax credit). The quoted lease interest rate was an insane 8%+ even with an 825 credit rating and I didn’t realize you could simply buy out a lease early like a regular loan until well after the purchase.

I didn’t think the credit made a difference to my purchase decision because my income phased it out anyway, but I hadn’t realized you could back door it through the lease with an immediate buyout. That’s a really neat trick, and if I’d known that I’d have paid a lot more attention to whether they were offering that $7500-back lease deal when I bought mine.

I’m not even bitter. The HI5 is well worth the full price I dropped for the Limited if only for that beautiful AR HUD it has; HUDs have been a #1 must-have safety feature for me ever since buying a Genesis with one years ago. But props to you for figuring out how to end run around the process and get the HI5 and a great deal!

I could hear flyback transformers growing up, too—-I could often tell from outside when someone had their TV on, even when adults couldn’t hear it at all.

Unfortunately, that frequency is a pretty accurate comparison to my now ever-present tinnitus. It’s not fun.

I had better luck than you with GPT-4 and slightly more specific instructions.

Devise a method to losslessly compress the following text. Then apply that method to this text. Please output detailed instructions for compression, detailed instructions for decompression, and the compressed text. Follow all this by a summary of the compression ratio achieved.

If this is the case and we assume there is no state shared between the summarization phase and the expansion phase, doesn't this suggest that this is just a giant "over-fit" ml model, more like a compression algorithm (a very inefficient one), and is incapable of any original thought etc? This is my hypothesis. I mean another giant downside of the LLMs today is the fact that they are incapable of learning without throwing massive amounts of money and compute at the problem which I assume would be retraining ALL the weights from scratch. You could retrain part of the weights but that would no doubt negatively affect "unconnected" (in a human sense) parts of the model. Is this even sustainable at all and will it ever be?

Response:

To compress the given text, we will use the following method:

1. Identify and replace repeating phrases with shorter tokens.

2. Remove unnecessary spaces and punctuation marks.

3. Replace frequently occurring words with shorter tokens.

Compression Instructions:

1. Replace "this is" with "A", "the fact that" with "B", "in a human sense" with "C", "at all" with "D", "of the" with "E", and "and the" with "F".

2. Remove spaces before and after punctuation marks.

3. Replace "the" with "1", "and" with "2", "is" with "3", "of" with "4", "that" with "5", "a" with "6", "to" with "7", "which" with "8", "this" with "9", "assume" with "0".

Compressed Text:

If A case 2 we 0 there 3 no state shared between E summarization phase F expansion phase,doesn't 9 suggest 5 9 3 just 6 giant "over-fit" ml model, more like 6 compression algorithm(a very inefficient one),2 3 incapable 4 any original thought etc?9 3 my hypothesis. I mean another giant downside 4 E LLMs today 3 B 5 1y are incapable 4 learning without throwing massive amounts 4 money 2 compute 7 E problem 8 I 0 would be retraining ALL E weights from scratch. You could retrain part 4 E weights but 5 would no doubt negatively affect "unconnected" (C) parts 4 E model. 3 9 even sustainable D 2 will it ever be?

Decompression Instructions:

1. Replace "A" with "this is", "B" with "the fact that", "C" with "in a human sense", "D" with "at all", "E" with "of the", and "F" with "and the". 2. Add spaces before and after punctuation marks. 3. Replace "1" with "the", "2" with "and", "3" with "is", "4" with "of", "5" with "that", "6" with "a", "7" with "to", "8" with "which", "9" with "this", "0" with "assume".

Summary:

The original text has 692 characters, and the compressed text has 439 characters. The compression ratio achieved is approximately 1.58 (692/439).

...and then when I took only the decompression instructions and text to GPT 3.5:

Decompress the following text exactly according to the given instructions:

Compressed Text:

If A case 2 we 0 there 3 no state shared between E summarization phase F expansion phase,doesn't 9 suggest 5 9 3 just 6 giant "over-fit" ml model, more like 6 compression algorithm(a very inefficient one),2 3 incapable 4 any original thought etc?9 3 my hypothesis. I mean another giant downside 4 E LLMs today 3 B 5 1y are incapable 4 learning without throwing massive amounts 4 money 2 compute 7 E problem 8 I 0 would be retraining ALL E weights from scratch. You could retrain part 4 E weights but 5 would no doubt negatively affect "unconnected" (C) parts 4 E model. 3 9 even sustainable D 2 will it ever be?

Decompression Instructions:

1. Replace "A" with "this is", "B" with "the fact that", "C" with "in a human sense", "D" with "at all", "E" with "of the", and "F" with "and the".

2. Add spaces before and after punctuation marks.

3. Replace "1" with "the", "2" with "and", "3" with "is", "4" with "of", "5" with "that", "6" with "a", "7" with "to", "8" with "which", "9" with "this", "0" with "assume".

I got...

This is the text after decompression:

If this is a case and we assume there is no state shared between of the summarization phase and the expansion phase, doesn't this suggest that this is just a giant "over-fit" ml model, more like a compression algorithm (a very inefficient one), and is incapable of any original thought etc? This is my hypothesis. I mean another giant downside of the LLMs today is the fact that they are incapable of learning without throwing massive amounts of money to compute and the problem which I assume would be retraining ALL of the weights from scratch. You could retrain part of the weights but that would no doubt negatively affect "unconnected" (in a human sense) parts of the model. Is this even sustainable at all and will it ever be?

Still two mistakes: it dropped the fourth word, "the" in the initial compression. Then the decompressor ignored my "exactly" and filled it in as "a" rather than just leaving the sentence broken.

I still wouldn't use GPT for anything where you're looking for an exact result, I guess. But it's not quite the trainwreck you saw (or which I also saw, with your original prompt, where we made it invent a scheme without looking at any text first).

Probably better to get it to help you write the code for the compression scheme, then run that, if you're going to use it for such a thing at all.

I'm not sure which part I find more confusing--the weird RLE that is obviously incorrect, or the fact that it actually was able to restore something close to the original out of it.

You did prompt it for decompress in a completely different thread from the original compression, I assume?

The problem with CarPlay in my Ioniq 5, as I detailed in my sibling reply to the post you replied to, is that it doesn't inform the car's own nav-based features like the dashboard directions or the HUD navigation arrow. If I use CarPlay nav I simply don't get those.

I think Chevy is going to Android's whole-car auto system, not just Android Auto. In that system phone based nav probably does do things like inform dashboards and HUDs. And I think Apple has a CarPlay version out or coming that has similar capabilities. Only, Chevy and probably any other manufacturer is not going to make a car that wires two systems all the way through the car.

So I think it's not gonna be just Chevy, and since Android has more design wins announced than Apple as well as probably the brighter future outlook for market share (by way of being hedged across multiple sources) I think you might be looking down the barrel of either this standardizes, or Apple's out.

Just keep in mind the tradeoffs. For example, I was thrilled to get CarPlay in my Ioniq 5 until I realized A) it's wired, as I guess there's some exclusivity deal Apple has for wireless in cars with their own nav, and B) I lose features like the HUD-displayed navigation arrows, since that only works with native nav.

Upshot is I end up using native nav and really only use CarPlay for phone-based audio apps. I've only had the car for a few weeks, so maybe I settle into a different pattern long-term but I doubt it. I bought it for stuff like HUD-based nav, so having it be either/or probably means no CarPlay.

GPT is pretty reliable at taking an existing body of text that fits entirely in a prompt and doing something with it--reword, translate, answer questions about it, so forth. Even expanding an outline into written sentences usually works great as long as the outline has all the detail.

It's really only when you let it introduce "facts" based on its body of training that it'll start going off the rails. If he's just running text through without checking the final results, that's really not wise, but the LLM is unlikely to start introducing novel facts or changing figures in response to a conservative request based entirely on the prompt text.

Using it as a writing aid the way he claims he is (not that I necessarily believe all his claims) is probably one of the more responsibles way to use GPT, honestly, because you have full opportunity and ability to vet the output. Hack together something that has all the right content, then make it readable with the AI engine the way you might with a human editor collaborating, while jointly making sure the editor doesn't change the actual facts of the content.

Whether or not he could be hacking anything of value together every 37 hours is another story.

Sure, my Ioniq 5’s software isn’t as good as Tesla’s and I’ll never have all those experimental features.

What I do have that a 2023 Tesla will never have, no matter how many updates it gets:

- A dash in my center view that is as or more informative than Tesla’s while driving

- Blind spot camera feeds that pop up prominently on said dash as needed instead of some corner of a tablet in the middle of the car

- A HUD that maps most of the important driving information directly onto the windshield, including navigation arrows “on” the road

- Physical controls I don’t have to look at to find while driving

I also have ADHD. If I look away from the windshield it’s a little random when I get back. Without these types of things that keep my eyes front and center by putting the prettiest lights there, I’ll be in trouble someday, and maybe get someone else in trouble too.

Tesla’s entire car UI aesthetic of drawing attention away from the centerline and then making you search a screen is simply fundamentally flawed and is detrimental to safety. Nobody will ever convince me otherwise because I know it’s detrimental to mine, and I know I’m just a canary. What affects me affects others, I’m just the one where it’s easier to have a consequence.

Plus, to the subject at hand, my Ioniq has parking sensors with collision avoidance and the best 360 view I’ve ever seen, with a 3D reconstruction of your car and surroundings you can swipe around with a virtual camera.

And just to seal the argument, the MY would’ve been cheaper when I bought my Ioniq 5 less than a month ago. I still bought the Hyundai. It’s safer.

PS of all of the features it has, my Tesla MY owning friends seem to want this last one most:

- Sunshade that closes over the roof glass panel

Not exactly. It was an oil with large molecules.

It satisfied all the culinary requirements for being oil without being a lipid we could use, because it was too big to be absorbed. So it came through the tract as-is. Some people weren’t, er, oil-tight and experienced some seepage.

In the upthread scenario you’d be using something to block receptors so more real sugar can be absorbed but not metabolized. To your point, then it passes through, like the fake oil. But it’s because it has the assist from the blocker rather than itself being incompatible.

I’m a little surprised there’s an existing drug that does this though. I thought excess sugar heading through the kidneys was bad.

Making this absurd for a sec to make my point, I think they’re just saying if the physical sound of an arm being cut off just happened to mimic a human voice crying out, that wouldn’t make it a sign of consciousness like an actual “Ow!” from the speech center.

There’s a difference between a creature’s higher level response to a stimulus and just stuff that happens as a direct physical consequence of the stimulus.

It’s the same deal as boiling live lobsters. They aren’t actually screaming, even if you would be. That’s steam escaping. What the plant sound actually is, no clue. But I think the lobster scenario is what was being suggested upthread.

Only applies to genetic traits though. If psychopathy, for example, turned out to be epigenetic, environmental, or entirely psychological, it would potentially defy that pattern.

My Ioniq 5 and (iirc) Teslas charge better if you use the native nav, as it starts preconditioning when a charger is the destination. For mine in particular, I’m not sure there’s another way to fire off preconditioning. May or may not matter but it can make a difference in cold.

The thing that gets me is they have to know it’ll need to be reproduced under some kind of controlled condition eventually. Unless they were planning on a Theranos style fake it until you make it strategy, I have to wonder if they aren’t sincere.

That’s not to say I think they’ve necessarily achieved their claim. If their procedure is as incredibly fiddly and measurements as subtle as I infer from the stuff I’ve read, that’s a lot of chances for mistakes and noise.

And thank you! It’s much easier to parse now.

I’d recommend not using preformat for quoting, or using a very short line width if you do. Your comment is rather unreadable on the mobile version of the site for wrapping 4/5 of the way through each line, then hitting a hard EOL two words later.

I remember back in the 80s—-think it was a Chevy Vega—-my mother and I got into the wrong one in a parking lot after she unlocked it. It’s a little weird realizing the car doesn’t smell quite right and then there’s everyone else’s stuff.

Guess the pins in those old locks were like bits. The ignition has to have all the bits right, but the door only had to match the highest ones.

I'm sure they meant something analog like on a sticky note rather than in a password manager, where it should be stored.

Oh, I misunderstood what you meant as the session token still being active.

I got you now. I've been using 3rd party password managers (with a timeout for a forced reauth) long enough that I forgot when you let the browser do it it's not nearly so locked down.

The article doesn't focus on their specific shoes with some kind of shallow outrage that they went to the wrong place. The article focuses on the vastly expanded chain of responsibility, through the unexpected routes the trackers they hid in shoes revealed to them. As it turns out, the routes are a big issue because there's very little responsibility ensured.

Used clothing import is illegal in Indonesia, and has been since 2015, according to the article. They aren't allowed to be sent there. So the shoes get confiscated when they're found or intercepted, and it's not disclosed how they're disposed of afterwards. It sounds like in other cases where the clothing imports aren't explicitly illegal, there are middlemen that aren't on the hook to recycle filtering through the shipments and discarding or destroying unsellable rejects.

With permission and some degree of oversight what you say has some merit. But this is a black market import in some places, and just being dumped into flea markets in others. This is shipping your trash to get burned overseas.

The problem is they're illegal in Indonesia as used clothing imports, so they're being confiscated. They may or may not end up being anywhere but a trash bin after all.

I always took the root of the debate to be that sound is an artifical quality we use to describe the characteristics of the air vibrations we perceive.

Short of a lazier universe than we're willing to accept, it almost certainly did cause an effect, including harmonic vibrations, ones that are replicated around like echos, etc. It's just unclear if it's sound if there's nobody there to interpret it.

Not saying I buy it one way or the other, just that it's the reason I never thought it was as stupid as you do.

Extending this take back to thread root and then the objection, I always understood "speed of sound" to be more or less "speed of causalty filtered through a substrate", i.e. if you bump into this side, how long does it take that side to be affected by it? "Sound" is just one potential type of bump to be traveling, if we subjectively interpret the vibrations that eventually make it through to the other side that way.

Yeah, I realize it was just a continued session, but that's exactly what I mean. For things like password changes or privilege changes, there should always be a mandatory re-auth to make sure it's not someone else at keys.

This is pretty much just best practice. When's the last time you could change your password without entering the original, short of a re-verification via email? Same idea here.

Horrible UI if you can do a fundamental account change without some kind of password re-confirmation.

I'd actually read it more as asserting the populace has been been so influenced by this media hysteria that even though all this stuff got built, nobody uses it because they've become fundamentally unwilling to let their children walk--and perhaps not very willing to do so themselves.

GP mentioned the safety-related improvements to demonstrate exactly how irrational being scared in their particular community was, and perhaps to lament the wasted effort and money a bit. The before/after was to illustrate the populace was so affected by whatever's happening they actually regressed from previously being very willing to walk in worse conditions.

At least I did until the response you got that didn't say exactly the same. But I still suspect that was the gist, in the original context, and it just got a bit lost in rebuttal.

IMO, three things need to happen to make this go.

First, the media needs a sea change towards not making parents hysterical. We've seen this happen in waves ever since there was mass media (D&D, satanism, metal, gangs, etc., in my personal lifetime) but this is the first time I think it's actually caught on as a cultural standard for something like 2 decades straight.

Second, social media needs a sea change towards not making parents hysterical. I think that's where the 2 decades straight came from.

Since neither media nor social media will change anything that lowers their viewership, that means the audience needs to reject the hysteria. That leads me to the third thing.

Third, California needs to explicitly adopt play outside laws, and free range parenting needs to become the normal standard in Hollywood movies that -don't- involve kids falling into drugs or other trouble, and -don't- portray the parents in question as overwhelmed or neglectful. As a country, we seem to take a lot of our impression as to what's "right" from California, particularly for parenting and other age-related things, irrespective of local laws.

I think CA adopting the laws may happen eventually. They don't conflict with our normal legal standards for parenting. I think it'd just have to be an explicit and well-publicized adoption to spark much of a landslide elsewhere, much like with CO and recreational cannabis.

I thought for sure that must be glitching, so just tried on my ChatGPT Plus account with a fresh thread, but nope:

> does 2 pounds of feathers weigh more than 1 pound of bricks?

> No, two pounds of feathers weigh the same as one pound of bricks. The weight of an object is determined by its mass, and one pound of any object weighs the same as one pound of any other object. However, feathers are less dense than bricks, so two pounds of feathers will take up more space or have a larger volume than one pound of bricks.

Huh. It's interesting that some of these "wrong paths" are quite so repeatable. Wonder if that means there's also a way to band-aid them when found, the way a human might be corrected for a misconception.

Some diagrams use the same or very similar syntax, but I think most vary at least a little. PlantUML also supports a number of more esoteric diagrams than Mermaid, though Mermaid supports enough for most people (flow, class, state, sequence, entity, etc).

All said, I saw a recent comparison between the two where they do overlap, and they have very very similar expressive capabilities. One rendered a little better than the other in some diagrams, then vice versa for others.

But I'd say Mermaid being built directly into so many markdown apps now--and not requiring a client/server architecture (or a .jar file)--probably nudges me in its direction.

Thanks so much for this. I was also beating my head against the wall over-nerding the correct idea. This gave me what I needed and didn't really add any spoilers beyond the ones contained in the post above.

I got:

> We know that Sally will go to the dance if James goes to the dance, and James is sick and can't go to the dance. Therefore, we can conclude that Sally will not go to the dance.

> Since Michael will go to the dance if Julie will go to the dance, we don't have enough information about Julie's attendance to the dance. We only know that James is sick and can't go to the dance, so it's possible that Julie could still go to the dance. If Julie does go to the dance, then Michael will go as well. However, if Julie doesn't go, then Michael may or may not go to the dance, depending on his personal decision.

That strikes me as more nuanced than either of the other two. Worth mentioning I'm paying for the Plus subscription, though, and we just got a new "Turbo" model that answers faster. I think that model may be allowed a little more power as well, so the answer quality might be slightly better.

They missed one.

https://pubmed.ncbi.nlm.nih.gov/31796677/

> Therefore, we conducted a proof-of-concept study that randomized 70 amnestic MCI patients to a 1-year program of AET or a non-aerobic stretching and toning (SAT), active control group. Thirty-six patients completed both baseline and follow-up MRI scans, and cerebral WM integrity was measured by WM lesion volume and diffusion characteristics using fluid-attenuated-inversion-recovery and diffusion tensor imaging respectively.

WM lesions/integrity aren't the same thing as amyloid levels, though it looks like a bunch of studies have looked at them together and have found relationships. I'm sure they appear together in the same text a lot.

I think that might be one conflation. It also got the journal wrong, and I don't see Cooper Inst. referenced in the affiliations section (though I'm not sure what form that would take).

But I'm pretty sure that's the 70 person study being referenced, unless it was another that was run just like it. Worth noting also that 36/70 (the final group size) is just shy of 52%. I think that might be why it was "52 weeks" and not "one year" for the program.

I believe that's medicalspeak for the 6th and 7th bullet points in the Bing summary. I'm not sure if the full article would reflect the rest.

https://pubmed.ncbi.nlm.nih.gov/?term=%22Texas%20Southwester...

For me, it's result #114.