dan_mctree

This seems to be a trend in general. Books or instructional videos are usually either made for people who are assumed to know basically nothing, or people who can be assumed to already know everything. I find it makes self guided learning very difficult in almost any field

Is anyone else surprised at how quick species can seem to evolve in these kinds of ways? Asian and African elephants are supposed to have separated millions of years ago, yet still strongly resemble one another. Yet in a few decades, something as fundamental as the presence of their tusks can shift dramatically. And that in a species with very long gaps between generations.

It makes me suspect there's something else going on here that's not just good old natural selection.

Have they done any tracking to confirm the tuskless elephants are actually being born to tuskless mothers? Or could it be that the parents are actually affecting the way their offspring are born in some lamarckian-esque sense? If there'd be some trick animals can pull to change their offspring in some ways, that ought to be highly effective for long term species survival and should thus be highly selected for.

Funnily the stated definition seems to fall for its own trap. Perception, here named as a particularly hard thing is something AI is starting to get pretty good at. Mobility lags behind a little, but we've seen some incredible demonstrations there too.

Perhaps it's always the things that are on our awareness horizon which seem hard but end up solvable. What's next? Perhaps long term planning, interference and on the fly learning? And then what are the difficult things beyond that horizon, the challenges we can't even see yet?

Why would they need melee? Just attach a lightweight single use gun barrel on top

Yep, we might face a world where gun drones have perfect accuracy from hundreds of meters away as well as split second kill on sight decision making. You'd be dead before you'd see it coming.

It's interesting that high p-values actually seem to more conclusively state something than low p values (like p < 0.05) do.

With a high p value, you can say with some degree of certainty that your test was unable to detect any effect. Whether it was due to the lack of an effect or because your test wasn't capable of measuring it

With a low p value, you don't actually really know if you detected something interesting. It could be due to a flawed test, biases, non-causal correlations, faulty p-hacky stats, etc.

So why do we consider the latter more worthwhile when it seems to say less?

Many of the common reasons that make companies able to pay more rent do not align well with preferences of locals. They could have higher prices, pay employees less, use lower quality ingredients, give less personal attention or externalize costs through say noise pollution. They also often have competitive advantages through brand recognition or company support, that don't actually increase quality

I figured it would be appropriate to ask an AI to come up with a label. So I prompted GPT-J with:

AI has proven to be terrible at moderating human content, we felt we needed a word for these kinds of buggy systems and have chosen: "

And GPT-J answered:

"LOLPOP"

Which seems to capture the spirit of unreliable AIs pretty well

Considering the Dutch connection. Thanksgiving may've just been a continuation of the Dutch dankdag/Thanksday, which is a small protestant celebration thanking God for the harvest. They do that every year, regardless of the circumstances, and it kind of sounds like they just continued that practice.

Harvest festivals in general were common throughout Europe and many immigrants would be familiar with them, so it's not very surprising that at least one version of it would survive.

A possible killer app might be something for business meetings. Videocalls are already a huge thing, but many feel like they're missing some real personal intimacy in there. If instead of a videocall, you could all be using AR to see each other walk around in front of you, that might make it a whole lot more appealing. If it's done right, you could have the same private one on one side conversations that are common in real life, but just in some AR space.

AR has a clear advantage over VR here as you wouldn't be blind to your real surroundings so you'd still be able to talk with the people in real life too (and not look like a doofus who walks into walls)

Even if that's the highest achievable, IQ tests designed to be usable for everyone tend to start getting really unreliable above 135 or so. Above that level, you're getting into the territory where test takers basically answer almost all of the reasonable questions right and where those few wrong answers necessarily start having much larger effects on the final score. Because of that, other issues start dominating the results. Random variance, sloppiness, mechanical issues in timed tests, how well you slept, whether there's slightly bad/ambiguous questions, whether you've seen that type of question before, whether you're good under pressure, cultural differences, etc.

For example, one serious official test I did included a timed test where you had to decode a string of symbols to text based on some key. I ended up being fast enough at decoding that I was effectively limited by how fast I could scribble down the decoded texts and turn the page back to the code, not by the mental calculations. I imagine others would have had similar issues, making this particular test at best capable of indicating a wide possible range of mental ability, a range which should be particularly inaccurate around the higher ranges as the ratio mentalTime/mechanicalTime decreases.

There also just aren't enough high IQ people to really effectively study the accuracy of these tests above those levels. There are no good baselines to compare these people and no real ways of checking whether the differences actually matter in any meaningful way.

It's generally referred to as day dreaming when people do step into their 'imagination devices'. Personally I do that quite a lot, especially in dead moments like on the bus. I find it to be more entertaining than using a mobile phone with terrible internet. Though, it's probably not amazing as you might imagine. For most people, visually imagining things isn't as vivid as real life, it lacks detail, and there aren't really any surprises in what you imagine. But most of all, if you're used to this 'power', it just feels mundane. Someone who can't walk might assume that walking is an amazingly pleasurable experience, but for someone who walks every day, it doesn't particularly feel very special.

And indeed, the contents of what I imagine are rarely something I talk about. It feels private, and like dreams, it's hard to convey the experience to others. You might also not recognize it if I do share something about my daydreams. Suppose I said: 'I was thinking of going to the beach', then you might assume I was thinking more along the lines of your typical thought experiences, while I was actively visualizing said beach.

That would make sense if all things are equal, but there are efficiency issues here. Having to take the extra step to convert the fossil fuels into electricity first (plus the later battery storage) means the same fossil fuels don't deliver as much power to the vehicles. That said, it's probably a wise move to move to electricity powered vehicles anyway because of the slow shift to reusables.

Yeah effectively it's rich people deciding that enormous sums of money should be invented out of nowhere and given to rich people. And somehow everyone is ok with that

Cars and bicycles are mostly equals in terms of who gets to get first. Though, in cities, it's indeed common for cars to take up a lower, more careful position and they will frequently let bikes go first even if the law may've said they were allowed to go first.

There's just so much stuff going on around you when you're driving a car in an inner city in the Netherlands, that it's wise to drive very carefully. In contrast, bikes have their own lanes, can use both sight and sound more easily and don't have to be so afraid of bumping into pedestrians, this means they have a far better grasp of the situation and possible risks. This combined with higher agility means they can move confidently and adjusted as needed.

Are there any hardware reqs to work with this?

Even diversified countries are vulnerable to similar shenanigans. There are a lot of vulnerable points in modern society. Destroy a couple of electricity substations or power line towers and you could paralyze cities housing millions.

These kinds of targets are also really hard to defend and comparatively easy to destroy. Hell, you could probably drive around tossing copper wires over power lines and inflict gigantic economic damage that way. (You would also probably die)

The first 10 for me were:

NYT, NYT, NYT, NYT, Vox, NYT, WaPo, NPR, CBS, LA Times

They certainly seem to like New York Times

Couldn't marine creatures learn to collect air bubbles underwater in order to use fire?

In addition to what the others said, stock trading fees often don't scale with the cost of these stocks. If you're investing with low amounts, these static fees eat up a lot of any potential gains of day trading

>This problem is accessible to anyone who knows basic high school algebra

Are you sure? Spotting the pattern isn't easy and I imagine a lot of high schoolers would give up in frustration while going down wrong roads. A good portion probably wouldn't even write down the numbers without errors. And even if you get past that, most high schoolers don't know how to do/write proofs. I'd be shocked if more than 10% could actually complete all your steps.

With challenging problem solving like that I'm afraid you're going to breed a lot of resentment.

Possibly the main benefit of democracy isn't that the crowd make wise voting decisions, it's that it makes politicians feel compelled to make wise decisions to save their careers.

I think you can't. The problem is that we don't actually see 3D, we basically see 2D twice. This gives our senses enough info to reconstruct a 3D mental image, but the information we gain isn't truly 3D.

Actual 3D-sight might include us seeing the inside of objects, what's behind them, etc.

We're aided in converting our 2x 2D sight into 3D mental models through the use of color and parallax. In projections on images, it's easy to just use color and get rid of parallax, and still get sort of close.

Suppose we tried to make a 3D projection of a 4D world, we don't actually get any significant new tools. We're still going to be limited by having to use parallax and color. This doesn't allow us to add enough information to add that 4D projection into 3D as neatly as the 3D into 2D works. At best you could try to distribute the information somehow. For example by giving information about the third dimension through brightness and the fourth dimension through hue. It certainly wouldn't work nearly as well though.

Commandline instructions fail as often as IDE based instructions do. Changes in different versions, changes in the environment, different ways of trying to use the software, different ways projects were set up; all of these and many other reasons cause plenty of issues with commandline based instructions. It's never "just do everything on the command-line", it's "it just spewed out an essay of errors on step 2, and google is silent"

Having to wait an hour for a download/install to complete is completely different from requiring an hour of browsing stackoverflow to figure out why the CLI install instructions given aren't working on your device.

For me it's the opposite. A modern IDE is significantly easier to understand, with far less mental baggage than some hacked together combination of several CLI based tools. The mental drain of clicking some nice green play arrow on screen when you want to execute code is non-existent. The baggage of remembering which tool does what, what weird flag you have to keep adding to one of the commands you need, what keyboard shortcut you needed again to exit your CLI based text editor, all of that takes far more baggage for me. Especially when you have 10 different setups like it, and may not return to some of them for months.

And even if you do somehow remember how all of your CLI tools fit together, you're still missing out on an incredible amount of power and clarity that IDEs provide. A CLI just doesn't have the ability to show information nearly as well as a GUI, it doesn't even come close in terms of the amount of bits worth of information you can show simultaneously, nor in its ability to use positioning and formatting to clarify what's important and related. Add that GUIs provide constant in your face hints that tell you what your options are, and it's no wonder almost everyone finds them significantly more pleasant to use. You don't have to remember anything, it shows you.

Yes, people end up developing bad habits and will run into some problems later. But these issues aren't as significant as people giving up early because of learning curves that are too steep on software which is only vaguely connected to what they actually want to do. "I want to build an app" is what drives people, not "I want to learn all the flags for javac".

Yeah if anything it's people moving away from IDEs that's making life much more difficult. The complicated part of an install is rarely ever just downloading some software, clicking the executable and then 'next' a few times.

The hard part is docker throwing out nonsensical error messages into your console, offering 3 equally illegible possible solutions, which after 2 hours of googling and attempts turn out to be wrong, because all you needed was a reboot to get to the next error (which ends up being caused because of one of the 3 attempts to fix the earlier problem).

Yeah sure, with an IDE you might not grasp what really happened, but do you really with the 15 step install instructions? Both GUIs and CLIs only ever show the information that the programmer decided to output anyway, and GUIs are actually much better at showing legible output due to all the extra space, formatting options, etc. So what is it that people think makes a CLI somehow grasp what's happening better? The fact you just copy pasted some command with 3 flags?

Cool little game! Loving the art and the morbid story. It was tough as nails

One nitpick: sometimes the game would seem to get unresponsive if I clicked the little hint door (it'd get stuck halfway open). Only spamclicking around for a while got me out of it

You can only really do customer support if it makes financial sense, which it won't unless you make a significant amount of money on your average customer. Tech companies that don't have sales, but instead take their revenue through ads or through selling data are making cents per customer. With average profit that low, even 1/1000 customers making use of your support for 5 minutes would destroy any chance of profit.

Comparing it to three random pictures of roads at sunset:

https://pixabay.com/p-280863/?no_redirect

https://pxhere.com/en/photo/118396

https://pxhere.com/en/photo/1165757

The main difference appears to be that the pixar picture is way blurrier, the lines and detail in reality are just way more crisp. Perhaps the pixar picture is supposed to be in motion, but motion blur should be reduced over distance while even the background trees and sky are blurry.

The blacks in pixars image also don't really seem black enough for a sharp sun at sunset. You'd think at least the cabels right in front of the sun would have higher contrast. While on the other hand, some highlights don't seem to be as strong as you expect (the metal railing on the opposite side)

Overall, you kind of expect your eyes to hurt in a situation like this, but it's all very soft and gentle.