truantbuick

Kevin Drum is an absolute shill for LCH. There was a study in New Zealand that, if you actually read the paper, unambiguously fails to support the LCH. What was the headline for the article he wrote about it? "New Zealand Study Provides More Support for Lead-Crime Hypothesis"

https://www.motherjones.com/kevin-drum/2017/12/new-zealand-s...

I was specifically responding to the commonly held belief that modified gravities and observations of the bullet cluster are absolutely irreconcilable. I'm not necessarily drawing any line in the sand about whether I believe in modified gravities over particle dark matter, or whether that distinction is even important.

In general, I'm suspicious of dark matter. It would make an interesting development if it ended up being the planet Vulcan of our time, where we doggedly postulate its existence despite mounting counter-evidence because it's unpleasant to question baser theories.

https://en.wikipedia.org/wiki/Vulcan_(hypothetical_planet)

First time I've heard of Fossil and started reading up on it: https://www.fossil-scm.org/xfer/doc/trunk/www/fossil-v-git.w...

In particular, I'm interested in the last section "4.2 Features found in Git but missing from Fossil".

Maybe this is some deficiency of my workflow, but both those things make Fossil sound extremely unappealing to me. You're telling me I have to push all local changes every time I want to push anything? What if I was just debugging or experimenting with one branch, and then I had to switch to a "serious" branch to push some critical changes?

And rebasing is fantastic when you have a feature branch workflow.

Some would disagree:

http://backreaction.blogspot.com/2018/04/no-that-galaxy-with...

As a non-scientist, it seems like dark matter has been relentlessly adjusted and overfitted and still appears very inadequate to fit all the evidence.

That article tries to debunk a slightly different notion than I had in mind. It's not that I imagined everybody was sitting just above 4 minutes when running the mile, and that the significance of those round numbers psyched everybody out until Bannister finally broke through.

Instead, it's about knowing what's possible and attainable. It's akin to the story of George Dantzig, who solved two unsolved problems in statistical theory because he believed them to be mundane homework problems.

Put it this way: how can something that had never been done before 1954 is now considering a very basic benchmark? I would argue that it's precisely because it is considered a basic benchmark that it is so common. If you know something can be done, that makes it easier to train to. If you know something is commonly done, then all the more.

Of course, that goes for any record/benchmark, but Bannister's 4 minute mile seems like a particularly significant one.

When we talk about something so big that it might as well be infinite, we call it "astronomical". We've all heard the awesome facts: we know we're an invisible dot in our town; we know the town is an invisible dot on the planet; we know the planet is an invisible dot in the solar system; we know the solar system is an invisible dot in the galaxy; we know the galaxy is an invisible dot in the universe.

But, as impressive as that all sounds, that's describing scale with exponential growth, but exponential growth pales in comparison to combinatoric growth.

The number of ways you can arrange 100 people is far greater than the number of estimated atoms in the universe. And not _just_ greater, greater in a way that is difficult to explain. A drop in a bucket? No. A drop in the ocean? No. All our metaphors are physical, and so they're inadequate, but let's just it's beyond all that.

Of course, that doesn't mean the probability of other intelligent life is best assumed to be of a combinatoric nature, but let's get ahead of ourselves and let our awe of the universe's size cloud our judgement. It's far from infinite, and silly to assume intelligent life is certain.

A few things that raised my brow:

1. The press release for the study and this article have completely opposite conclusions. At least the author is relatively upfront about this, but he characterizes it as if the press release is just a cautious, sciencey take on the results. By the way, it's not just the press release, but the abstract/paper has pretty unambiguous language saying findings "failed" to support such hypotheses.

2. There's a Mother Jones chart that uses data from the study, but it's cherry picked. They take a few cohorts, ignore the (very valid) adjustment the study authors made, and the chart itself is IMO misleading because when you scan it, you're likely to mush up the highs and lows of each cohort to make it look more linear than it is.

3. Looking this guy up, he's clearly an advocate around the lead crime hypothesis, so it should be understood this guy has an angle.

So, yeah, this headline sucks.

What the graph really seems to indicate is that time is only linear when working within a cache size on the author's computer (remember that iterating a linked list accounts for the gradual increase in between the cache jumps). If the theoretical upper bound of RAM access was really the important factor at this scale, I wouldn't expect it to be almost flat and to suddenly jerk up every time we have to go to the next cache.

Assuming the author's O(sqrt(n)) is correct, it seems only relevant on much, much larger scales.

In light of that, it really doesn't make sense to pollute the typical use of Big O notation. It should always be understood to be just one metric to understand an algorithm.

Clearly there are exceptions. In my area, I don't come across lights that use sensors (or at least the sensors pick up cyclists).

I'm just uneasy with saying "red lights = stop signs for bicyclists". In complicated intersections, you need to wait for the green, unless it's very obvious there's no oncoming traffic.

Cyclist commuter here. Red lights should be red lights to everybody.

However, Idaho stops are the way future. It's just doesn't make sense to apply the same "come to a complete stop" standard to cyclists. They're traveling at a much lower speed and have much less ability to accelerate. The extra time it takes to come to a complete stop (which means finding your balance at a stand still), and then remounting the bike accelerating through the intersection is not worth it, and makes the commute that much more of a slog.

I think characterizing the Idaho stop as "rolling through the intersection" might be a little bit misleading, since it conveys a really lax attitude. Idaho stops are generally quite safe.

Edit: In most cases, Idaho stops have very similar cadence to a car coming to a full stop. You slow down up to the line, look all around you, determine the right of way, and if you have it, you start peddling back up to cruising speed. The fact you don't come to a complete stop is pretty much a technicality because the timing is almost the same.

If you come to a complete stop, the timing is all off, and there's much more likely to be hesitation and ambiguity. Starting from a stop takes a lot more time, is much more likely to lead to false-starts, and puts you in a very vulnerable position.

So are drugs.

I'm not sure I agree with the idea that this somehow characterizes the idea of quicksort better.

Technically, you can always convert a recursive algorithm to iterative and back again (and not just with existence arguments -- it can be generally constructed), so why pick on quicksort specifically?

The only thing achieved here is putting the partitioning of the list and sorting in two separable parts. Whether that characterizes quicksort any better is a matter of opinion, I reckon.

I'm sure I saw a HN link a month or two ago touting the great success of Munich's Linux move. Can anybody find it?

Sorry, is the vim that ships with OS X consider emulated?

SPE is an absolute mess as an experiment, and shouldn't really be regarded as science.

However, you really don't even have to go far so far into the technical details to find it to be flawed. The whole premise seems absurd to me.

You get some students, tell some of them to act like prison guards and the rest to act like prisoners, and then it's supposed to be surprising that they follow through on this?

"Well, we didn't expect them to be so sadistic!"

There's nothing I read that was more sadistic than a fairly typical hazing ritual. All participants knew this was ultimately voluntary and could quit at any time.

Using these kind of game metaphors just shallowly rationalizes something of far more importance, IMO. Legal or not, a country was worse off because of his direct and deliberate actions.

I don't know that much about finance, so can somebody explain to me why I never hear anybody censuring Soros over this? I only hear congratulations for bamboozling the incompetent British government.

He made billions at the expense of an entire country.

I'm not saying it was illegal, and maybe people like Soros are needed to help deter governments from artificially manipulating their currency like the British did. However, when it comes down to it, he benefited by executing a bet meant to enrich himself, compound the problems of the pound, and diminish the wealth of a nation. It seems like a plainly immoral thing to me.

I didn't imply otherwise.

I just didn't explicitly mention global variables because that will always remain a possibility that lurks over everything. Technically, we can never rule them out because it's possible that all the universe and all the events of universe were preordained by one massive global variable, without any particular rhyme or reason for rules and laws for mechanics of local behavior.

Lots of errors in this article. It particularly rankles my sensibilities when he says randomness doesn't actually exist and it's just when humans can't track all the variables.

Randomness is much more interesting than that, and while it would make a lot more sense for the sake of classical mechanics, it cannot possibly be explained away so neatly according to Bell's Inequality.

FAQs with categories often annoy me. Maybe I've just had bad luck or I have a selective memory, but it just doesn't seem to work that well when I've used them. I understand that they're often necessary depending on the volume of content, but I very much prefer just reading down one simple page even if tends toward the long.

It's reasonably quick to scan, or use CTRL-f to look for keywords (without worrying that they might be on another page), and often times simple single-paged FAQs are written better and more cohesively.

Furthermore, even if I'm scanning over a question I wasn't specifically looking for, it answers other questions I might have, gives me a broader picture of whatever I'm trying to do, and informs me of subjects I hadn't even thought to consider.

There's a lot of difficulty in evaluating stuff like this, and I feel like we inevitably trip over biases that go in both directions. On one hand, it's very easy in hindsight to ridicule what didn't come true or what seems like an outdated 1960s infatuation. However, it's also easy to pick through his predictions and pick through the modern day and apparently find a match.

Ultimately, I give some credit that he wasn't completely wrong (which I suspect is a lot harder than it seems!), but I don't think these predictions were particularly visionary -- just intelligently made given the information of his own time, which is all we can really expect.

He reasonably charts the technology of his own time to today. Some of it turns out alright, some of it not, and some in between.

I understand this is a bit of fun trying to rethink a very old thing, but the benefits are really bare even when looking past all the damage of a potential transition.

Basically, you're consolidating the unevenness between years/months/days of the week to one special 1 or 2 day period. Ok, that's alright, but it still requires a bit of extra help. You still have to think through calendars understanding "Oh! I have to add a day for the epagomenal. Oh, it's a leap year, so it will be 2 actually."

That might be preferable to having to think through 31 or 30 or 28 or 29 days, but not by that much. Is aligning months and days of the week to the same time every year a good thing? I suppose it's easier to determine what day Thanksgiving is every year, but otherwise, it seems like a minute benefit. On the other hand, I wonder if there are some benefits of that not happening. The only one that comes to mind is it would suck for your Birthday to be on a Monday every year, but I suspect there might be other reasons where enforcing a day to occur on the same place in the week could be harmful.

I'm no expert myself, but I thought the Turing-Church thesis implied the exact opposite. Turing machines and lambda calculus are widely assumed -- though not proved -- to be the limits of what can be computed. And yet, in a Gödelian fashion, for a given computer (equiv. set of finite axioms), you can always formulate deterministically true statements that cannot be computed.

Note that the formulation of these are counter-examples and thus assert a fundamental limit of computation, but it says nothing that there might be many other things that are non-computable that do not fit the form of these counter-examples.

There's something that particularly bothers me about Goodfellow's article. He condescends about crazy theories while making ridiculous assumption after ridiculous assumption while acting very authoritative about each one. Just because his theory is less "sexy" (it doesn't involve hijacking or terrorists or suicide) doesn't make it any more valid. I'm appreciative of commentary that urges a more prudent approach, not a condescending and equally implausible theory that happens to be boring.

Real life has never sprung a random topic on me at 7:30 AM and required me to have a handwritten essay written 25 minutes later. This essay, in a wildly naive attempt to limit the subjective value of writing, was judged largely on "coherency".

We had computers back in 2005, and almost all students typed their school essays out, but the college board hadn't got with it. Since it took me so long to handwrite things, I didn't have a moment to think about the topic. Instead, my strategy was to prepare an extremely formulaic essay, and then furiously write as fast as I could. Hey, it worked, except I was already stressed and tired by the end of the essay with over 3 hours of test to go.

To backtrack a bit, I've always assumed the SAT to be about aptitude for reasoning, not about throwing "real life" challenges at you.

As a member of the first year of students who were required to do the writing section, I'm glad it's gone. What an absolute crock that was, and it probably screwed up the other sections as well. When you make an already ponderous test over an hour longer, it becomes less about aptitude and more about endurance.

This has probably been addressed millions of times, but just to give a response to why this is a misleading comic:

If you were equally mocking of both Bayesian and Frequentist, the Bayesian would arguably come up with basically the same conclusion. In this scenario where the Frequentist ignores previous data based on human history and our understanding of the lifecycle of the sun based on physics, the Bayesian should do the same if we are fair. Then his prior distribution would likely be 50% belief that the sun would explode (Why favor either outcome? Thus the prior distribution is split down the middle between both outcomes.). With the new information given from the detector, his posterior would be 35/36 probability that the sun has exploded.

Maybe the Bayesian approach provides a more explicit way of incorporating previous information, and without that, it's cause for some misuse with a Frequentist approach, but that doesn't mean Frequentists need be fundamentally ignorant in their approach.

The comic's good for a laugh. Just don't take it too seriously as a criticism.

While almost everybody agrees that pure rote memorization is not usually the right approach in (any kind of) teaching, I tend also to be very skeptical of the oppositely pure approach.

We might reason that the programming languages we usually use today are not likely to be the ones used tomorrow, but the tremendous amount of understanding one can achieve using the languages of today can still be useful tomorrow.

Notation/code are purely conventions -- the idea they can express can be expressed in any number of ways. Nevertheless, ignoring notation/code can be harmful. Paraphrasing Leibniz, notation can be the difference since the necessary labor of thought is wonderfully diminished.

My guess is that most of these "miracle technologies" won't work out the way we hope, but that doesn't mean it isn't interesting to track their progress. Plus the research being done on them might still be useful. Ultimately, there might be some property of graphene that makes it unsuitable for widespread use, but some other related material might slide into its expectations.

Unfortunately, we get attached to certain big names and lay all our hopes and dreams on their backs, and they so often don't work out. We are only human, and we can't follow the whole of all technology that lies on the horizon, but even if the big names don't work out, there's still a lot of promise in this area, and I get excited by the research.

I can't fault Airbnb hosts if they use this -- if I was a host, I probably would. Nevertheless, it takes some of the charm and advantage out of it for us hostees. Airbnb seemed like a marketplace where sellers for the most part aren't trying to optimize their profits to squeeze out every last buck.

Edit: Actually, to be more accurate, it didn't feel like much of a "marketplace" at all, and that's where the charm was. You weren't terribly aware of supply and demand and the rest of econ 101 when using it.