Comments by "Chrysippus" (@4.0.4) on "Computerphile" channel.

  1. He proves he knows about the subject but gives us zero knowledge. That was the point right?
    629
  2. I think it's really important people understand 256 bits is not twice the security of 128 bits. It's twice that of "255 bits".
    475
  3. CNN and BBC is True News™
    47
  4. You can trust NerdVPN, where we drink military-grade coffee.
    43
  5. [MUFFLED OY VEY IN THE DISTANCE]
    39
  6. So this means you can simulate the sequential memory of a Turing machine using only functions and parameters?
    26
  7. Buzzword soup: "data mining big data with AI-based cloud computing"
    22
  8. I saw someone implement turing completeness with checkboxes and CSS. It was slow as molasses.
    20
  9. This is a really great idea, and branding private, local "clouds" that do not send your personal info anywhere but at the same time bring greater functionality as "Web 3.0" is a great way to make it clear that this is the way forward.
    18
  10. I wonder how many IBM 360s would "fit" into a Raspberry Pi Zero, either with emulation or with (optimized) recompilation for ARM. "if it ain't broke"... oh boy.
    16
  11. KARASIRA That's called winning. Being self-employed and not failing is harder, riskier and much more profitable and enjoyable. You only work at a company if you can't build one instead. Stress and thrill are two sides of the same coin.
    16
  12. Come on, Linus Tech Tips is just clickbait that delivers, and that's why it's popular. Also computer science is not as popular as "consumer geek culture".
    16
  13. Python for me. The language is a perfectly acceptable 6/10, but the tools for it are a Rube Goldberg machine where you kinda need docker because otherwise "it works on my machine!"
    15
  14. That's not a bad video but the premise is misleading. You're not talking about companies, you're talking about top labs who can snob the plebian developers because they pay top money for top talent. A regular company (95%+) will not look for "comb-shape" talent (or rather, they'll have to settle for way less).
    12
  15. I never thought the difference between "man" and "woman" could be so intertwined in a language! I can imagine it would make some people uneasy.
    8
  16. As someone who actually used it, it's very barebones. Like, I wanted custom easing curves for transitions and it was easier to code that in AviSynth and render from that. I expected some GUI for easing curves in Blender, linear fades look really robotic.
    7
  17. Or in pre-SP2 Windows XP there was a "feature" to send system pop-ups via the network. They looked weird, but it was enough to scare someone.
    7
  18. We call him Poem Billy!
    7
  19. I think cleaning is about getting rid or filling in data, and reduction is about summarizing, think the map-reduce stuff.
    7
  20.  @khashayarr  not really. Just give it a bunch of examples of API calls and corresponding requests from users.
    6
  21. I don't think we have a problem with compatibility really, as standards have evolved to a great extent. You can even run a program written for one OS in another. But nothing is cleanly designed and it leads to heaps of abstraction. One one hand this also means skilled programmers can pull off amazing tricks in intensive tasks like video codecs and game engines. x264 is a work of art.
    5
  22. I never imagined how clever this was! I can only imagine this is used for more than just faces right?
    5
  23. avts I wish all science videos had CC, it makes it easy for translators. That said ze English of ze German zientist iz fine. It's a cute accent!
    4
  24. I think the biggest problem here is companies rarely want to really be decentralized, and without giving up control it's just a marketing gimmick, not really decentralized.
    4
  25. And that can be demonstrably (statistically) fairer (more likely to predict if you'll pay back your debt or not) than any human who decides based on emotion.
    4
  26. I was curious about how you train these. Could be fun doing it ourselves even if the results are crappy/undertrained.
    4
  27. ¡My favorite X-Hombre is Carlos!
    4
  28. It is viable to make a chocolate with 20% or less cocoa. Just ask Nestlé.
    4
  29. Due to the recent waves of Islamic migrant terror in Europe, I think we should ban TOR, and any calculation that involves big prime numbers (processors could do a check and throw a hardware exception if big prime numbers are detected).
    3
  30. I've seen prompts that make ChatGPT write in such a way that every single AI detector (well, 5 of them) marks it as 100% human.
    3
  31.  @JustinJFain  thank you for your response! And what an interesting-sounding job title.
    3
  32. I noticed that too lol You just have to type sin(x) on whatever to get a proper plot.
    3
  33. ​ @marcosolo6491 he'd have to read the name twice, to confirm it was a Fluke.
    3
  34. It's always going to be a human way of handling it, because no software can know what you care about in the data or what's really relevant on it. Depending on how you "fill in" the missing attributes, you might skew the conclusions.
    3
  35. Just pause and think what you can do now with an ESP8266 board that runs at 80MHz, does WiFi, a lot of chip peripherals, has onboard storage of 4MiB, a ton of IO, and costs less than $3. A guy on YouTube named CNLohr even made it output 3D meshes in NTSC video, and run a Minecraft server.
    3
  36. I really want a video just on Support Vector Machines! (Example: why would a traditional neural network outperform it?)
    3
  37. smalltime0 You still can get that random device driver to phone home like he said - basically negating the usefulness of firewalls.
    2
  38. So wait, is D-Wave a bunch of snake oil?
    2
  39. AnavrinSM64 I thought there was only one kind of quantum computers ("qbits" going through "quantum logic gates", somehow). So Google just bought it to R&D and it's not used for anything yet?
    2
  40. This! Also even mobile readers feel so much less convenient than scrolling through a webpage. PDF is good for printing but the strict control it gives to the publisher/creator means that any reader software fights an uphill battle to make it less "accurate" and more convenient instead. PDF was built around presentation, as opposed to other formats built around semantics.
    2
  41. One thing he forgot to mention was the whole issue with CompuServe's patents (one of the reasons PNG took off so nicely).
    2
  42. Not all emulation is the same though, some instruction sets are almost subsets of others - this is why you can emulate ARM on x86 with much higher efficiency than x86 on ARM - "transpilation" sometimes works
    2
  43. It's amazing how MP3 (with the amazing LAME codec) can rival much newer formats just because of better psy models, kinda like x264 rivals H.265 codecs for the same reasons. Still, this particular video was LAME for different reasons...
    2
  44. There are many many algorithms, K-means is just one, there's a free plugin for image editors called XiQuantizer that has many algorithms like neuquant, which is neural-network based. There are also simpler algorithms than K-means, after all people had to make gifs a couple of decades ago lol
    2
  45. Do actual terrorists really use Facebook and WhatsApp? I don't buy it. And if they do why would we raise awareness and not just catch them for once?
    2
  46. The most surprising thing is that it shows promise of becoming better than this with a larger dataset. You know this is going to happen at some point!
    2
  47. If you don't mind me asking, what do you use R for? Do you happen to know/use general-purpose language(s)?
    2
  48. Can you guys please train a large language model with Miles so I can keep asking him questions?
    2
  49. I really thought it would be some new exploit, but I learned something new about network infrastructure instead. What can I say... ... Acknowledged?
    2
  50. This and Miles (the AI guy) are my favorites.
    2
  51. This is just stupid. It's like saying 1+2+3+4..+ ∞ = -1/12. Yeah, on some level, on some weird definition of it, on an impractical world of mathematicians that love weird logic, it's true. But if you can run through all branches of conditionals in a program why couldn't you brute force this problem? I know that "echo Hello World" halts. It can be proven, for this particular program, and I could write a program that proves that this program halts. PRAGMATICITY.
    1
  52. Blender's video editor is also pretty usable.
    1
  53. If you're worried about the singularity, fear not, we've been "solving" fields like computer vision since the sixties, and look what we still can't do at the end of Moore's law.
    1
  54. Wasn't it shit already?
    1
  55. ProfDaveB There's something about you that can turn a mind-numbing Wikipedia article into cozy time with Grandpa Dave. Congrats!
    1
  56. So basically it can't be fixed? If it only uses documented behavior a lot of programs would stop working if it's fixed, right?
    1
  57. hellterminator I'm pretty sure there's undocumented features out there that Microsoft not only keeps but emulates for backwards compatibility. This could well become an exception, but "undocumented features" are commonly relied upon when people need a quick hacky solution.
    1
  58. hellterminator But Microsoft does get hands dirty with virus protection. Not only their Defender product, but also the Malware Removal Tool that's issued in Windows Updates. I think the fact that this bug got a name and press coverage could push them to do it.
    1
  59. By the way, I downloaded the 150-million Adobe user's file, just to check. My password was only used by ONE other person! Yay! One of my friends had a password used by more than a hundered people. You should check it out.
    1
  60. I don't understand. The FSA has a memory of one word, in this case up to 25 and worth say, 5 bits. What if its "single word" of memory was worth a trillion bits? And, it had a ridiculously large but technically finite graph of all the operations it could perform on that single word (or number) composed of a trillion bits? (In other words, couldn't you create a Turing complete machine with an absurdly complicated FSA?) As an example, he says you can't have the machine remember which coins led to 15, but adding a few bits to the number you could store that using a bit of math (mangling the number but it would still be extractable by a simple modulo).
    1
  61. I don't understand. The FSA has a memory of one word, in this case up to 25 and worth say, 5 bits. What if its "single word" of memory was worth a trillion bits? And, it had a ridiculously large but technically finite graph of all the operations it could perform on that single word (or number) composed of a trillion bits? (In other words, couldn't you create a Turing complete machine with an absurdly complicated FSA?) As an example, he says you can't have the machine remember which coins led to 15, but adding a few bits to the number you could store that using a bit of math (mangling the number but it would still be extractable by a simple modulo).
    1
  62. As someone who doesn't do Matlab or R but was wanting to pick R (R studio looks nice), R and Matlab should be similar in terms of performance, right? R seems more programmer-oriented.
    1
  63. I don't understand. The FSA has a memory of one word, in this case up to 25 and worth say, 5 bits. What if its "single word" of memory was worth a trillion bits? And, it had a ridiculously large but technically finite graph of all the operations it could perform on that single word (or number) composed of a trillion bits? (In other words, couldn't you create a Turing complete machine with an absurdly complicated FSA?)
    1
  64. +Billy Sugger and booleans are stored as 4 bytes in many high level languages
    1
  65. How could you possibly ever design a "super-intelligence" and know that it isn't inherently flawed in some way? General intelligence is not a series of if-statements.
    1
  66. I'm sorry but is any website really still vulnerable to this? It's hard to believe. Isn't everyone using some sort of MVC framework that has an object-relational mapper built in?
    1
  67. Actually it's pretty normal for GIFs to have one palette per frame. It's usually sub-optimal though.
    1
  68. I'm disappointed, the guy tells you to use salted hashes. Use f*king bcrypt if possible!
    1
  69. Any time a function calls itself you're using recursion. It's not uncommon...
    1
  70. Adam Brown Arduino is not a processor, it's a board with a microcontroller on it.
    1
  71. No, please, you don't need that. The best way is to use bcrypt (look it up), but making a salted hash (look up that) is easy and most people use it.
    1
  72. I would love this too, hope they consider it. Especially the more clever stuff, like stuff out of code golfing, or even historical clever code like quake's iconic square root optimization.
    1
  73. Haskell sure is a "computer science" language, because coding like that sure won't be efficient for anything but proofs of concept.
    1
  74.  @dafoex  "do I really need data privacy if I have nothing to hide?"
    1
  75. I had to do a double take to make sure this video wasn't from 2018.
    1
  76.  @codeman99-dev  conda: solving environment for over a dozen hours before I cancel
    1
  77. Can this read and generate MIDI? I wonder which styles it fails/works on.
    1
  78. There was a time when EXIF metadata wasn't removed by social media. It's a similar "bug".
    1
  79. Will the speed of light be a limit to clock cycles? Is it already? (Say, for how far the RAM is from the CPU, copper traces, etc) I ask this because a fraction of a thousandth of a millionth of a light-second is in the ballpark of just a few centimeters.
    1
  80.  @mytech6779  Thanks, that cleared things up. If it was a problem for chip design it would have already been long ago then, but it's a bit mind-boggling to imagine a CPU cycle "propagating" through the chip and computer like water pressure through pipes rather than happening "simultaneously".
    1
  81.  @igorthelight  sorry to disappoint you but quantum entanglement specifically cannot be used to transfer information. Not even gravity travels faster than light, nothing does. The only real workaround is to bend spacetime like an Alcubierre Drive does.
    1
  82. Whoever bought Ryzen must be pretty jovial... No performance hit. For Intel, check 4K reads on an SSD before and after patching. And buy AMD next time lol
    1
  83. 0:31 - "It drinks in C program statements and it spits out at the other end [...]" That's not how it works at the other end.
    1
  84. Worse is that he speaks Spanish, which is very similar to Portuguese (Lua is moon in Portuguese).
    1
  85. I really don't understand reducing a game's quality far below the lowest setting (there are retro games for that!), but I admire the dedication.
    1
  86. Why do people talk so much about the Y combinator? It seems a problem on the language itself, for not having syntactic sugar for recursion. If you do it in this strange manner, it's the language's fault for not having it implemented clearly.
    1
  87. Nikolay Yakimov that's cute, but it doesn't mean anything in the majority of cases. You need to have the ability to define a function in terms of "2 copies of itself". It's just semantics. What I'm asking is how is this of interest to anyone? It's just imposing a contrived constraint and "solving it" with a method that only works in a specific language that for some reason can make copies of itself but you don't call it recursion.
    1
  88. This is great content. It genuinely makes me want to pick RStudio and try to learn data analysis.
    1
  89. Fingerprint sensors are great for usability and practicality, but they are a bit of a security nightmare. You can copy the fingerprint from a photo, this has been done.
    1
  90. The sine wave is a sin. Get it? Ok I'll leave
    1
  91. What about using two discriminators for different (but similar) things with the same generator, and in the noise input, using one point in the noise as an indicator of what you want? (The generator's input is all noise except one value) Then later, you could have a slider where you decide how much of either you want. You could even use face recognition to train it on known facial features and landmarks.
    1
  92. ".....nooo.... we should not be worried....." Sounded really honest thanks. I'll take your word for it and forget that little Snowden thing.
    1
  93. I wonder, aren't Kinect cameras very cheap right now? Does anyone know how hard it is to use several?
    1
  94. This is the most didactic content I've ever seen on the channel. Great stuff!
    1