Greg's recent activity

  1. Comment on Animating URLs with Javascript and Emojis in ~tech

    Greg Link Parent
    Apparently the history API is rate-limited, so this trick was using the old location.hash = technique instead - sadly we may just have to live with boring non-animated URLs for at least a little...

    Apparently the history API is rate-limited, so this trick was using the old location.hash = technique instead - sadly we may just have to live with boring non-animated URLs for at least a little longer!

    2 votes
  2. Comment on Animating URLs with Javascript and Emojis in ~tech

    Greg Link
    I clicked through, dutifully chuckled at the concept, enjoyed the video example, and then hit the back button to comment here. And then hit the back button again. And again. I guess it's fairly...

    I clicked through, dutifully chuckled at the concept, enjoyed the video example, and then hit the back button to comment here. And then hit the back button again. And again.

    I guess it's fairly obvious in retrospect, but mutating the URL many times per second will pretty much destroy the browser history's usefulness in short order!

    15 votes
  3. Comment on Why airlines make flights longer on purpose in ~misc

    Greg Link Parent
    True, but only assuming the fuel is correctly priced. Right now the extraction and refinement is reflected in the price, but much of the cost relating to emissions is covered communally (and in a...

    If it was more economical (fuel) for airliners to be propelled by something other than the jet engines while on the ground, we'd probably have seen some adoption of this at some level.

    True, but only assuming the fuel is correctly priced. Right now the extraction and refinement is reflected in the price, but much of the cost relating to emissions is covered communally (and in a way that disproportionately affects certain geographies but disregards national borders). It's effectively a subsidy to anyone using fossil fuels.

    I'm a big proponent of market-based solutions, but in order to make them work here there needs to be something along the lines of a universal carbon tax that's high enough per kg of CO2 to sufficiently cover a global climate indemnity fund.

    3 votes
  4. Comment on A Bill Banning Most Abortions Becomes Law In Ohio in ~news

    Greg Link Parent
    This all clicked into place for me when a friend pointed out that these people would do (and are doing) exactly that if the roles were reversed - they know this and don't want to be on the...

    This all clicked into place for me when a friend pointed out that these people would do (and are doing) exactly that if the roles were reversed - they know this and don't want to be on the receiving end. It's like a horrible bastardised version of the golden rule.

    5 votes
  5. Comment on The key to glorifying a questionable diet? Be a tech bro and call it 'biohacking' in ~health

    Greg Link Parent
    The comment about "a cow" is obviously just a bit of hyperbole for effect, but everything in that paragraph does suggest he's eating fewer than 1,000 calories. I don't think it's fair to criticise...

    The comment about "a cow" is obviously just a bit of hyperbole for effect, but everything in that paragraph does suggest he's eating fewer than 1,000 calories. I don't think it's fair to criticise her understanding of nutrition just because you didn't like the literary choice to use a bit of exaggeration.

    “fish, chicken, or some steak,” plus arugula, spinach or “sometimes asparagus or Brussels sprouts” and finally, “I have mixed berries as a dessert.”

    Context suggests that anyone who eats like that certainly wouldn't describe a pretty substantial ribeye just as "some steak", and even if he did, the other two options are even lower calorie. Steamed or grilled green vegetables add next to nothing there, and a decent portion of berries adds another few hundred. One meal a day on that diet, without very unusually large portions or dousing the entire thing in olive oil, is absolutely less than 1,000 calories.

    2 votes
  6. Comment on Mathematicians Discover a More Efficient Way to Multiply Large Numbers in ~science

    Greg (edited ) Link Parent
    Just for clarity, there are four algorithms mentioned here: "Grade school" long multiplication: O(n2) Karatsuba: O(nlog23), which is roughly O(n1.58) Schönhage-Strassen: O(n logn log logn) The...

    Just for clarity, there are four algorithms mentioned here:

    • "Grade school" long multiplication: O(n2)
    • Karatsuba: O(nlog23), which is roughly O(n1.58)
    • Schönhage-Strassen: O(n logn log logn)
    • The as-yet unnamed new one from this paper: O(n logn)

    Before this point we knew that O(n logn) was theoretically possible, but not how to do so - the breakthrough here is that the authors have now figured out how to achieve it in practice.

    If the algorithm is only useful for really large numbers, I would assume that means it's actually inefficient for numbers below a certain size?

    Absolutely correct. If you look at the Java BigInteger source, for example, it gives implementations for all of the first three that I mentioned above to use for different sized inputs.

    But if so, what is that minimum size?

    In practice, there's an informed but arbitrary line drawn based on the number of digits - it depends on the exact implementation, the architecture, and potentially a few other things as well when it comes to actual wall clock execution time. In the enormously unlikely event that you're the person who needs to derive these numbers, you might start with an order of magnitude estimate from theory, but you'll almost definitely refine it empirically by running some real world benchmarks and seeing where the lines cross over - which is why the comments in that Java library are peppered with "This value is found experimentally to work well".

    And do the conclusions that there are algorithms for n^1.58 and n*log(n) theoretically apply to all calculations where n>=2?

    Yes, they do. And with this new paper, they aren't just theoretical conclusions - there's now a working example that we can use!

    To understand why we have the threshold issues even though any of the algorithms will work at any scale, it's important to remember that big O notation is only concerned with growth, not with size. A simple example to hammer this home would be that an algorithm that takes 100,000,000,000,000n operations is O(n), but one that takes exactly n2 operations is O(n2). You can see that the latter grows much more steeply, but that doesn't inherently make it slower because the base size of the former is large; unless you're regularly dealing with n>100,000,000,000,000 it'll be faster to use the second option, but order notation isn't the tool to tell you that. Quickest and easiest way to figure it out? Eyeball the code for any obvious starting points, find out there probably aren't any, and then do the real world benchmarking I mentioned above.

    Hopefully all that made some kind of sense! I'm a little rusty on all of this, so if anything looks off then point it out - it could well be my fault.

    [Edit] Fixed the first mistake I spotted! The 100,000,000,000,000n crossover point is at n>100,000,000,000,000 not n>10,000,000 as I originally put. That would be the case only if it were O(1) complexity with a fixed cost of 100,000,000,000,000 operations. I also now regret typing so many zeros when making a silly enormous number because it's making this whole thing hard to read.

    11 votes
  7. Comment on Mathematicians Discover a More Efficient Way to Multiply Large Numbers in ~science

    Greg Link Parent
    The wording of the article isn't especially clear, but from what I can see in the paper they've got an actual algorithm that achieves the previously theoretical lower limit of O(n logn), compared...

    The wording of the article isn't especially clear, but from what I can see in the paper they've got an actual algorithm that achieves the previously theoretical lower limit of O(n logn), compared to the Schönhage-Strassen algorithm's O(n logn log logn).

    The article confuses matters a bit by talking about the improvement from a naïve O(n2) method to Schönhage-Strassen (which is very significant but, as you say, old news) rather than the tiny (but mathematically interesting) difference that this new algorithm brings.

    11 votes
  8. Comment on Move back to your dying hometown. Unless you can’t. in ~life

    Greg Link Parent
    I find this absolutely fascinating, because your concerns sound so much like my own, yet your comfort is the exact opposite. I'm not for a second trying to debate or dispute, it just really...

    I find this absolutely fascinating, because your concerns sound so much like my own, yet your comfort is the exact opposite. I'm not for a second trying to debate or dispute, it just really resonated even though it's so different.

    For me, the city is the only place where I don't feel the need to perform. I never need to worry about how people see me because I know they aren't even looking at me. The people I care about are always there, close by when we want to spend time together, but I can still choose to be anonymous when I want - another face in the crowd, rather than the focal point on a quiet street. The background noise, both literal and metaphorical, is hugely comforting.

    And then when I do want to go to a concert, or a restaurant, or a park, or anything else, they're all there and waiting for me. Every day there are a thousand things happening, so whether I choose to take part or not, I don't ever need to worry about missing out because there'll be a thousand more tomorrow.

    That's what city life is for me: the place that I can take a deep breath and relax, because none of it is depending on me.

    10 votes
  9. Comment on “Just Switch to Linux” Is The Loser’s Game in ~tech

    Greg Link Parent
    It's pretty smooth when it works, but I'd describe my experience with Wine as a little bit like running web apps in, say, Internet Explorer 9 or 10: things generally function, and sometimes...

    It's pretty smooth when it works, but I'd describe my experience with Wine as a little bit like running web apps in, say, Internet Explorer 9 or 10: things generally function, and sometimes they're perfect, but they haven't really been tested for that specific environment so you're a lot more likely to hit visual quirks, non-responsive buttons, or general weird behaviour.

    1 vote
  10. Comment on “Just Switch to Linux” Is The Loser’s Game in ~tech

    Greg Link Parent
    You can run it in a browser, similar to Google docs, although I don't think it's quite at feature or speed parity with the native versions. There are definitely MS Office power users who need all...

    You can run it in a browser, similar to Google docs, although I don't think it's quite at feature or speed parity with the native versions. There are definitely MS Office power users who need all the features and plugins to do their jobs and can't switch (and I'm of the strong belief that for certain use cases Excel is the one MS product that's actually superior to all of its competitors), but I'd say that at least 90% of people are just riding on familiarity and compatibility at this point - they're all good candidates for running it on a Chromebook or similar.

    12 votes
  11. Comment on Survey: 83% of US teens have an iPhone, Android 9% in ~tech

    Greg Link Parent
    That's true, and I agree with you that it was a missed opportunity for them, but for whatever reason it doesn't seem to have hindered WeChat, WhatsApp, or Snapchat.

    That's true, and I agree with you that it was a missed opportunity for them, but for whatever reason it doesn't seem to have hindered WeChat, WhatsApp, or Snapchat.

    1 vote
  12. Comment on Survey: 83% of US teens have an iPhone, Android 9% in ~tech

    Greg (edited ) Link Parent
    In that case all this green bubble dread makes a bit more sense to me! Still seems optimal to just sidestep it with one of the many cross platform apps, but I guess it doesn't feel that easy when...

    In that case all this green bubble dread makes a bit more sense to me! Still seems optimal to just sidestep it with one of the many cross platform apps, but I guess it doesn't feel that easy when you've got to tell everyone else in the group to do so too.

    [Edit] Thinking further, I wonder if Apple originally had this affect the whole group just as the easiest and most consistent technical solution, or whether it was deliberately made to gently peer pressure non-iOS users. I'd guess probably the former, but it could go either way.

    3 votes
  13. Comment on Survey: 83% of US teens have an iPhone, Android 9% in ~tech

    Greg Link Parent
    I doubt they see value in competing when they've already lost. I wouldn't want to go back to SMS/MMS either, but with the notable exceptions of the US and China, pretty much everyone else (on both...

    I'm completely dumbfounded that Google seems wholly uninterested in competing with it

    I doubt they see value in competing when they've already lost. I wouldn't want to go back to SMS/MMS either, but with the notable exceptions of the US and China, pretty much everyone else (on both platforms) uses WhatsApp already.

    4 votes
  14. Comment on Survey: 83% of US teens have an iPhone, Android 9% in ~tech

    Greg Link Parent
    Does it affect the whole group, or just the member on Android? I'd been assuming the latter, but if it drops everyone in the group to MMS that'd be more than enough reason for me to use a...

    Does it affect the whole group, or just the member on Android? I'd been assuming the latter, but if it drops everyone in the group to MMS that'd be more than enough reason for me to use a different client to talk to them!

    1 vote
  15. Comment on Survey: 83% of US teens have an iPhone, Android 9% in ~tech

    Greg Link Parent
    I hadn't internalised what an important business move it was for Apple to integrate the messaging client into the OS. When you said "network effects", I didn't initially get how a phone OS could...

    I hadn't internalised what an important business move it was for Apple to integrate the messaging client into the OS. When you said "network effects", I didn't initially get how a phone OS could even have a network effect because I'm in a WhatsApp dominated market, so your friends' OS choices don't matter.

    Smart of them to integrate it (because now Apple, Snap and Facebook are competing on the messaging field where Google have been left out), and smart of them to provide only SMS as a fallback because it works enough not to leave people out but still kind of sucks enough to push them towards the client all their contacts are using.

    5 votes
  16. Comment on Should I go to college for computer science? in ~talk

    Greg Link Parent
    It's a cliff, so that definitely could be an issue, but two years is a short enough period that I imagine it's probably easier to take the hit and know it's only temporary rather than artificially...

    It's a cliff, so that definitely could be an issue, but two years is a short enough period that I imagine it's probably easier to take the hit and know it's only temporary rather than artificially cap oneself and then scramble to negotiate a raise in month 24.

    There's also the social expectation of reciprocity which probably helps on the whole - not enough to rely on solely, I'm sure, but a lot of people will feel an obligation to pay back what they "owe" to an organisation that they worked closely with for almost a year.

    3 votes
  17. Comment on Should I go to college for computer science? in ~talk

    Greg Link Parent
    Can't say I have, although the words "income share agreement" have been floating by in conversations a few times over the last year or so, so I think we'll be seeing how the model works out in...

    Can't say I have, although the words "income share agreement" have been floating by in conversations a few times over the last year or so, so I think we'll be seeing how the model works out in general over the next couple of years.

    I'd say I'm cautiously optimistic about it as long as the upper and lower limits are fair (as Lambda's seem to be), but it wouldn't take much for an unscrupulous organisation to make it quite onerous.

    3 votes
  18. Comment on Laziness Does Not Exist in ~misc

    Greg Link
    I'm concerned that the author doubles down on the absolute "laziness does not exist" point at the start and end of the article - it's not just clickbait for the title, it seems to be a genuine...

    I'm concerned that the author doubles down on the absolute "laziness does not exist" point at the start and end of the article - it's not just clickbait for the title, it seems to be a genuine belief, and I honestly think it's a damaging one.

    The bulk of the article, the point that often the appearance of laziness is actually due to much deeper problems, is important and valuable. It was well put and I hope more people internalise it, because it should be the base case approach for dealing with anybody who might be struggling.

    I'm speaking from far more experience than I would like when I say I know there are times that getting out of bed really isn't an option. I know intimately what mental health problems, and the treatment, and medication, and all the time and stress and side effects those bring do to a person's ability to function. But I also know how easy it is to convince yourself that today is one of the impossible days when actually it's just a difficult one.

    I'm constantly looking over my own shoulder to make sure I only use it as an "excuse" (sometimes to others, more importantly to myself) when I genuinely have to, not just when sitting around doing nothing seems more appealing. If laziness really didn't exist, that'd be carte blanche to take the easy path a lot more often, and I think I'd be a lot worse off for that.

    9 votes
  19. Comment on Should I go to college for computer science? in ~talk

    Greg Link Parent
    Feasible? Certainly, but it's quite a different skill to coding. I know for a fact I'd make a much worse sysadmin than I do a developer, for instance, so at the very least I'd suggest keeping an...

    Feasible? Certainly, but it's quite a different skill to coding. I know for a fact I'd make a much worse sysadmin than I do a developer, for instance, so at the very least I'd suggest keeping an open mind and a willingness to be guided by which you find yourself better suited to.

    In terms of value, I'd call it a higher risk option. Everyone needs devs, from the occasional freelancer at a tiny company to technical leads of thousand person departments at Google. Everyone roughly recognises what a developer does; they might still think that a six-month project can be done in a week, but they at least grasp why you're there and worth paying.

    Sysadmins, on the other hand, only make sense as a dedicated hire at very large scale - they'll be managing many machines, potentially for hundreds of clients, so the absolute job pool is much smaller. Often there will be developers or IT staff doing some of the duties as an ad hoc part of their job, with varying levels of success, when a dedicated hire would be overkill. It's also often a rather unappreciated position: a corporate manager sees a high salary, a job that "surely one of the other techies can do", and no tangible output; when you're doing the job well, you're invisible ("so what are we paying you for?"), and when you're visible, it's because something went wrong ("so what are we paying you for?!").

    Don't get me wrong, it's a hugely interesting, important, and often well paid field - and maybe it'll be perfect for you - but one of the reasons I'm appreciative of those who do it is because I know I'd suck at it.

    5 votes
  20. Comment on Should I go to college for computer science? in ~talk

    Greg Link Parent
    Very much this. I haven't had occasion to work with any real hardcore network engineers, but I can say without hesitation that a good sysadmin is worth their weight in gold. Unless your company is...

    Very much this. I haven't had occasion to work with any real hardcore network engineers, but I can say without hesitation that a good sysadmin is worth their weight in gold. Unless your company is huge you might not need them in-house, but you're still relying on some of the best at AWS to keep things running on your behalf. As far as I'm concerned they're the keepers of dark and arcane knowledge that allows the rest of us to run our code in quiet, reliable comfort.

    3 votes