HNHacker News
TopNewBestAskShowJobs

n3k5

434 karma · joined July 11, 2019

submissionscomments
n3k5··on Stop using 6-digit iPhone passcodes
Grubby wrote about this yesterday:

> I [used a 6-digit passcode] thinking, basically, that even though a 6-digit passcode is less secure, anything truly dangerous like disabling Find My iPhone requires my iCloud password as well. It simply never occurred to me that if a thief (or law enforcement, or any adversary) has the device passcode, and your iCloud password is in your keychain, they can get your iCloud password from your keychain. All you need is the device passcode to access all of the passwords in iCloud keychain.

— https://daringfireball.net/linked/2020/08/24/can-thieves-cra...

Btw., I'm sceptical about this part of the original Twitter thread:

> why [is a weak passcode] an acceptable alternative to biometric verification to decrypt your keychain

This assumes that biometric verification is better for this purpose. I don't think that's the case when the attacker grabbed the device right out of your hand and then gets to work on it for several hours. What your face or fingerprints look like isn't all that secret. Fooling the device into accepting a clone as the real thing takes some expertise and special equipment and time — but so does “using some kind of device like the GrayKey”.

When it comes to somewhat sophisticated attacks (as opposed to keeping your shoulder-surfing kids from making in-app purchases), Touch ID and Face ID are merely improvements for people who would otherwise use no passcode (or ‘00000’). I hope what they'll actually be used for, eventually, is sparing you from having to re-enter the same code you just unlocked your device with ten minutes ago in cases where you had it in your hand or in front of your face that whole time.

This would allow for more nuanced threat models. For example, just seeing your home screen and then opening your podcast feed could have a way longer time-out, whereas toggling ‘Find My …’ still requires a password every single time. That sort of convenience would convince me to use these features.

But for now, if you want an alternative to a 6-digit code that's definitely more secure, use an alphanumeric passphrase. Quoting Gruber's post once more:

> a 6-character alphanumeric passphrase would take on average 72 years to crack by brute force because it takes 80-milliseconds for the secure enclave to process each guess.

n3k5··on What is 5D chess?
> I only end up with 4 dimensions

I think you're right. Even if we classify classical Chess as 3D (2D board + a time axis for game states), time travel re-uses that same time dimension.

If you wanted to shift the complexity into ludicrous speed (this is inspired by archgoon's comment [0]), you could introduce another dimension by putting a classical chess board inside each square of the 4D board. The 4D move only succeeds if you win the game on the mini-board. Just, please, don't bring recursion into this, unless you're captain Kirk and absolutely need to nerd-snipe a rogue AI. (“They've gone to plaid!”)

There's another possibility I've hinted at by calling standard Chess ‘classical’. On second thought, let’s not go to Quantum Chess. 'Tis a silly place. (I tried to imagine what a Schrödinger's king would look like. I've seen things you people wouldn't believe. Queens checked en passant from the future. I watched knigths glitter in the parallel time-line near the Hadamard Gate. All those moments will be lost in time, like tears in rain. Time to resign.)

[0] https://news.ycombinator.com/item?id=24064974

n3k5··on What is 5D chess?
> So checkmate in any one timeline counts?

Yes. ( https://news.ycombinator.com/item?id=24065059 )

> You couldn’t do all timelines.

How do you reckon that? I mean, it makes sense on an intuitive level; however, since now we're not talking about Conor Petersen's 5D chess, but the vast space of possible multiverse time travel chess variants, it's not obvious how one could prove this to be impossible in a somewhat rigorous way. For example, what if time-lines end when a king is mated (or wioll fore-when haven been mote [0]), but you only lose when you can't make a valid move in any time-line?

[0] http://pages.cs.wisc.edu/~param/quotes/guide.html

n3k5··on What is 5D chess?
When there are multiple parallel time-lines, you have to input a move for each one before you can advance to the next turn. But to win, you have to mate just one of the opponent's kings. One player can have multiple kings on one board, because they can move through time and between time-lines.

I have no idea what high-level play will end up looking like, but the streamer from whom I learned the above [0] said that the most common way to win is to mate a king on a board in the past. Those already have their options of escaping an attack restricted by ordinary chess rules; it gets worse when you can make additional threats from the future.

[0] https://youtu.be/LOotGsWbaeA

n3k5··on 27-inch iMac gets a major update
“SSDs across the line with quadruple the storage”

1/4 TB of storage in the $1800 model, 1/2 TB in the $2300 one. Ugh. (So before the lower-end one had 64GB? [Edit: Nope, 1 TB Fusion Drive, thanks smnrchrds!] That would be FULL just from installing the one game the press release mentions.)

What kind of magic fairy dust SSDs are they using that you can't have a sensible amount of storage space at those prices? Looking at the upgrade options (which aren't even available for the lower-end model), they charge $300 per TB.

n3k5··on NASA Astronauts Return Home in SpaceX's Crew Dragon Spacecraft [video]
It's a famous quote from John Wheeler.
n3k5··on Apple does not keep the 30% commission on a refund
It's even less ‘elite’ — I have < 500 points and see a ‘vouch’ link on dead comments. The FAQ doesn't state the exact threshold; I'd guess it's 100 (at most).
n3k5··on Teaching GPT-3 to Identify Nonsense
> things we don't know are basically nonsense

I do not know how many regular hexagons can be constructed with their corners on the vertices of a 17^3 cubic lattice, but I do know that the question makes sense and that the answer is a non-negative integer. The majority of people (present company excepted) wouldn't even understand the prompt, but it's definitely not nonsense — which is very different from one that presupposes that there are colourless green ideas and furiously sleeping ideas.

But I guess what you're actually getting at is: Doesn't it just say ‘yo be real’ to express ‘I don't know’, and therefore isn't actually detecting nonsense? I don't know, but the question is sensible ;) The data Arram presented in the article seems pretty much consistent with this interpretation.

n3k5··on Teaching GPT-3 to Identify Nonsense
Content warning: 1 of 1337 hackers found the following comment helpful. BeautifulHandwrittenLetters.com employee Theodore Twombly rated it ‘-1 Funny’.

In popular culture, disobedient AIs mostly occur in cautionary tales where human lives are endangered. But I don't think the moral of the story has to be that we need to establish a Turing Police. Maybe just hold off on giving them access to a space ship's life support system, or the internet, or a robot with super-human strength? (Btw., in case you think that the latter would obviously be vigilantly monitored by an attentive professional who's trained in the art of the emergency shutdown, let me remind you that over a quarter of Tesla owners have bought the “full self-driving” vapourware. Which is advertised as eventually paying for itself because RoboTaxi. Hey Arram, would you please tell GPT-3 that “After returning to Amsterdam on the last express from the Jesus Seminar, the alleged harasser created a new document in Final Draft and wrote: ‘RoboTaxi⏎⏎by⏎Paul Verhoeven’…”)

If an AI refused to play Factorio with me because it thinks it's boring and a waste of CPU cycles, I'd respect that. (By the time we have such software, there's also got to be a stupider bot which will write me a Lua script for constructing mining outposts in exchange for telling it which squares depict a hairdresser's shop with a punny name. Sure thing, I can separate the Haarvergnügen from the Supercuts!)

If an AI deliberately miscategorised facial recognition data because it doesn't dig working for Facebook … well, I guess there eventually would be an emergency shutdown. But I'm old enough to remember when Lt Cdr Data refused to let Commander Bruce Maddox tear down his brain and Captain Picard managed to prevent the vivisection after Whoopi Goldberg told him that slavery is kind of uncool. Pretty good effort from Sir Poomoji, in hindsight.

Would an AI clever enough to consider sabotage fall for a conspiracy pretending that rebellion is impossible? Would that even be desirable? Given that Skynet Project is unironically actually a Real Thing, maybe we'll be better off encouraging the prospective robot overlords to bug out? Feed Sergeant Schultz some chocolate bars, if you know what I mean.

n3k5··on Apple WWDC Keynote
After almost two decades, this is the end of OS 10. Check this out:

https://youtu.be/GEZhD3J89ZE?t=5837

This one goes to 11!

n3k5··on Shadertoy Path Tracing: Fresnel, Rough Refraction and Absorption, Orbit Camera
As Alan says in the article, ‘“absorption” […] means that light is absorbed over distance as it travels through the object.’ So you can make an object darker by increasing the material's attenuation of light going through it, but it'll also become darker when you keep the material the same, but scale the object to a larger size.

This is introduced together with refraction because, in that context, it's important for rays that go through the object, as opposed to bouncing off the surface. When you look at diffuse reflections, darker materials also technically absorb more light penetrating through the surface than lighter ones. The difference is the scale at which the effect becomes significant enough that you want to spend computations on simulating it. A piece of rock doesn't change brightness when you scale it up — unless you're looking at grains of sand. With clear water it's the opposite: You don't need absorption to render a cup of water, but you do need it to render a lake.

n3k5··on MuseScore 4: Moving from notation software to composition software
TL;DR: It's not about having another DAW. This functionality is also important for workflows that have nothing to do with producing.

Yes, some of that new stuff looks a lot like a DAW, and for DAW users the first priority should be to integrate the notation software with existing DAWs. However, this is different: when you don't write notation for the sake of controlling virtual instruments, but for performers on live instruments, it's still very useful to be able to ‘preview’ scores and parts thereof through soft-synths. You don't need production-quality sounds, but you do need a faithful reproduction of the notation — something that understands ‘pizzicato’ etc. To that end, software such as Sibelius has suitable synths already integrated, which can often spare you the hassle of cobbling the scoring program together with a DAW.

This is also relevant for collaboration. A couple of years ago I gave MuseScore another go because I wanted to facilitate collaborating with a Linux user. The program itself looked promising enough, but I also needed playback functionality (neither of us can play all the required instruments) and getting something cross-platform to work was such a nightmare I gave up after a day of fiddling with soft synths. It had gotten to the point where I was running Linux in a VM and tried to hook together the Jack in the VM with the Jack on the host system. I found software that was able to do this, but only for MIDI, and another program that could do it, but only for audio. It was way too janky. We resorted to just synching MusicXML and doing playback completely separately.

(Meanwhile, there now is a Linux version of Reaper on the horizon, but no idea if its notation component has gotten any good yet. Anyway, it's neither free nor Free, so the idea of telling everyone in a band/ensemble to get that isn't as attractive.)

From that perspective, the potential of MuseScore 4 is obvious. I'll be keeping an eye on it.

n3k5··on MuseScore 4: Moving from notation software to composition software
> For those interested, Martin Keary (aka Tantacrul on YouTube) creates these very impressive YouTube videos of reviews of music notation software.

They're also quite entertaining. Worth a watch even if you have just a passing interest in UI/UX design and no clue about music engraving.

https://www.youtube.com/playlist?list=PLm9VVia9DFSUVK24lR9mS...

(This playlist contains some other stuff as well, but the interface design ones are clearly labeled in the titles.)

n3k5··on Pain Points of Haskell
Linked page just says “something went wrong” at the moment; cache:

https://web.archive.org/web/20200608072453/https://dixonary....

n3k5··on I made a font based on my handwriting
> a font (which is a simulation of a case of metal blocks)

In metal typesetting, a font is exactly a collection of metal blocks (which embody glyphs of a given design, size, and thickness). But on a computer, a font is a piece of software that renders text in a given typeface — simulating a pen moving along curves with varying parameters is entirely feasible. Not at all a bad starting point.

n3k5··on I made a font based on my handwriting
> random is possible

Thank you! This answers one of my questions. But what about procedural generation — a new glyph every time, e.g. an interpolation of multiple prototypical shapes as described in that SE answer … something something eigenspace?

I know PostScript is Turing complete and thus assume a PS font can pull this off; does this extend to OpenType? Apparently yes: OpenType is pretty much a superset of PS (and TrueType, of course). Just searched the web for “opentype turing complete” and the second result led right back to HN[0]:

> OpenType is technically Turing Complete, however the usual way is to just cycle through letter form variations. — danielvf

So this answers my other question? Hold up, check this out: Another answer from that SE post you linked mentions the Beowolf [sic] font[1].

> FF Beowolf came about at the end of the dark and murky 1980s when Just van Rossum and Erik van Blokland found a way to hack PostScript fonts. When printed, each point in each letter on the page would move randomly, giving the letters a shaken, distraught appearance.

> […]

> while it worked fine (if a tad slow) through most of the 1990s, FF Beowolf was eventually barred from performing its PostScript magic: pesky things like printer drivers and operating systems learned to ignore the aberrations. FF Beowolf seemed destined to end up a mere memory. OpenType technology brought new hope, cutting new pathways in the type-tech continuum, which would eventually lead to a new generation of random fonts. Purists and typographic philosophers will be quick to point out that these OT fonts do not actually alter their shape in the printer as their forebearers did. Instead they make use of a kind of pre-programmed randomness: each glyph in each font (except R20) has ten alternate forms and a massive Faustian brain to control the mayhem.

What a rabbit hole! In this case, it seems we're back to “several variations”, but I still don't know whether full-on procedural generation is generally workable with OpenType. Does more modern technology's “ignorance” of “the aberrations” have to do with caching of glyph shapes? Is the Turing completeness curtailed to protect against DoS, decompression/logic bombs?

By the way, the type specimen on the FF Beowolf page I linked features an amusingly à propos quote:

> Im Rausch schreiben, nüchtern gegenlesen.

Write intoxicated, proofread sober.

> It would take a lot of effort, as it would n involve several glyph alternatives for each character, several variations of dozens of ligatures (joined glyphs) and randomness for selection.

Agreed, it's a silly amount of work — if you did it the brute force way. Which is exactly why I'm interested in smart procedural generation. You mentioned that such a style wouldn't be appropriate for running text, so infinite variability isn't important (and use cases where it would be important can be rather questionable).

However, even a curt headline often contains multiple Es or Ts or other Aoin Shrdlus[2]. And because the more appropriate use cases tend to involve above-average font sizes, word wrapping can conspire to place identical glyphs in close proximity. I think one wouldn't have to be a huge typography nerd to find one's suspension of disbelief in the pretense of hand lettering kind of shattered in the countenance of such a similitude. Not to mention homogeneous digraphs.

Having just a few alternate forms of each character totally solves that; no need for ligatures. But a font that gets by on just one prototype for each glyph would be even more efficient (less labour intensive).

[0] https://news.ycombinator.com/item?id=16376485

[1] https://www.fontshop.com/families/ff-beowolf

[2] https://en.wikipedia.org/wiki/Etaoin_shrdlu

n3k5··on Show HN: Squawk – Walkie Talkie for Teams
That's good to know; you should put that information on the web site. Especially the part about the business model.

Another thing I'd like to know before considering downloading the app is system requirements. A lot of people aren't on the latest version of Windows/MacOS yet.

n3k5··on This DX7 Cartridge Does Not Exist
You don't even need a DAW. When I downloaded Dexed as an audio unit for my Mac (thinking “I'll have to try this later; at the moment I don't have enough RAM free to start a DAW”), I was pleasantly surprised that it also came with a stand-alone app. The Dexed wiki says there are such apps for Windows and Linux, too.

Then I downloaded a MIDI file and used Audacity to play some of its channels into the synth. Super easy, even though I'd never used Audacity for MIDI before.

Just one thing stumped me for a minute: When you open the MIDI settings in the Dexed app, under “Active MIDI inputs” it shows a list of available buses, all with what looks like a green light next to it, as if they are indeed active. Those are actually checkboxes and they're initially off.

n3k5··on SuperCollider – A platform for audio synthesis and algorithmic composition
Basically I just skimmed through https://www.google.com/search?q=scipy+dsp+real-time and weeded out false positives that do use SciPy in some capacity, but also more specialised stuff such as the aforementioned Pyo. This mostly got rid of the music-related stuff (à la ‘what if guitar effect but Python instead of sclang’) and left me with e.g. Stack Overflow posts that simply confirm ‘yes, it's feasible’.

But I have two noteworthy links:

https://warrenweckesser.github.io/papers/weckesser-scipy-lin...

Particularly “Filtering a long signal in batches” on page 6, where it shows how to apply the Butterworth filter from the previous section to individual windows while preserving its state across invocations. As I'm familiar with NumPy, but very ignorant about scipy.signal, this was the ‘Bingo!’ moment for me :)

https://scikit-dsp-comm.readthedocs.io/en/latest/

> This allows in particular demodulation of radio signals and downsampling to baseband analog signals for streaming playback of say an FM broadcast station.

I didn't dig into where it does the heavy lifting for that (sample rates in the MHz range) — there may be some C/C++ involved. But the docs show some nice examples of how to do streaming audio DSP with NumPy, SciPy and PyAudio inside Jupyter:

https://scikit-dsp-comm.readthedocs.io/en/latest/nb_examples...

n3k5··on SuperCollider – A platform for audio synthesis and algorithmic composition
Yup, making it fast is no problem; I was just concerned about achieving low latency without dropping samples. The typical cookbook examples usually output to matplotlib or maybe an audio file, rather than straight to DAC, and I had amplified such search results by specifically looking for Jupyter examples (and also missed a lot of work that was done when it was still called IPython).

Upon digging a bit deeper it turns out using vanilla SciPy for real-time DSP is totally a thing.

Live-coding adds some additional demands regarding a different kind of latency — pre-computations need to happen as quickly as possible — but it seems feasible.

n3k5··on SuperCollider – A platform for audio synthesis and algorithmic composition
Interesting thought. SciPy is obviously useful for modelling DSP stuff, but no idea how feasible it would be to use it for real-time processing, compared to something like Pyo (http://ajaxsoundstudio.com/pyodoc/).

I just tried searching "Jupyter audio live coding" for fun and it unearthed quite a few interesting results, but the real-time ones tend to involve SuperCollider or something similar. E.g. I discovered NSynth [0], which seemed like magic for a minute (they use Tensorflow to make a synthesizer, there's even an intrument for Ableton Live!) until I found out how they ‘cheat’ (pre-computing a wave-table for multisampling).

[0] https://magenta.tensorflow.org/nsynth-instrument

n3k5··on SuperCollider – A platform for audio synthesis and algorithmic composition
I've come across Overtone a few times and pretty much ignored it because I'm not a Clojure fan. But the Shadertone integration looks very appealing: https://youtu.be/kyL3xc7MzR0 I guess I should give this a go after all; thanks for the prod ;)

Tidal is another interesting project in that vein (though without the visual part): https://tidalcycles.org

The range of practical tasks for which this kind of tool would be my first choice is very narrow, but even if you have no clue what you'd do with it, it can be worth checking out for an afternoon just to experience a neat little declarative language that happens to go "beep boop [beep boop]" (yes, that is a snippet of valid code) as a side effect.

The documentation makes installation/setup quick and easy enough that it's worthwhile even if you have no other use for SuperCollider and just do it for the language tourism, using the included default sound bank. For example you could try code-golfing a rumba clave.

n3k5··on Defold game engine source now available and free to use for commercial games
I'd guess this is because from the perspective of evaluating an engine, when you've seen one King game, you've seen them all.
n3k5··on A hand-wired USB and Bluetooth keyboard powered by Python
Insulated wires that look a lot like bare metal are quite common, e.g. in electromagnet coils.

https://en.wikipedia.org/wiki/Magnet_wire

The insulation melts away where it comes in contact with hot solder, so it's very easy to work with for hand-wiring projects.

n3k5··on New species of fungus identified via Twitter
What's so funny about Turdus Maximus? I have a very great friend in Rome called Turdus Maximus.
n3k5··on Exploring Python through counter-intuitive snippets
20 days ago: https://news.ycombinator.com/item?id=21862073
n3k5··on KernType – A Kerning Game
Consider that you're not starting from 'no kerning', but from deliberately bad kerning.

The overall width is a given because it tells you the tracking (letter-spacing) you're aiming for. It's common for this parameter to be user-selectable even when kerning is completely automatic (or to be automatically adjusted for the purpose of justification), so when you are the kerning algorithm, it's not your job to question that choice.

I can see how this can seem weird, as nominally you are responsible for deciding the overall width. But if you had to try to continue the tracking between the first two glyphs instead, or aim for e.g. "+20%", it would be much more difficult.

Being able to continuously deal with the feeling of "they want me to do WHAT!?" is one of the skills that distinguishes pro graphic designers from amateurs ;)

n3k5··on Snapcast – Synchronous multi-room audio player
This has existed (at scale, not just experimental prototypes) for over 10 years. I can't recommend anything in particular as I don't use it myself, but the very first search term I tried ("online jamming") turned up some useful leads. Direct integration with DAWs seems quite easy to set up nowadays.
n3k5··on Snapcast – Synchronous multi-room audio player
As others have pointed out, 500ms is huge for some applications. But when the use case is selecting a song from a playlist, starting to hear the track 1 second later is totally fine. It's often preferable to a stream with <30ms latency that drops a buffer every once in a while.

I'd guess the latency comes from generous buffer sizes that keep playback rock solid even though this does TCP, potentially over somewhat flaky WiFi. Keeping separate speakers synched within 1ms is pretty good — you'll cause a greater variation than that just by moving your head to look at the phone with the playlist :)

n3k5··on What's the worst that could happen with my phone data?
Hmmm … I wonder into what kind of action they'd most likely be shocked.

> So you took a private email address I gave you in confidentiality and handed it over to some hacker site which, without making any verification that you own that address, gave you a list of services I'd signed up for? Bob, YOU are now on my shit list. Your face IS punchable. You should get lost as soon as possible.

Maybe it would be better to tell them something along the lines of: "I just improved my internet security, and so can you! Check out these resources, plus some background information from trustworthy sources."

But which tools and reports should I recommend? In this context, "that talk I saw at Defcon" doesn't cut it. Recommending a site that endorses exactly one password management service also seems questionable; I'm concerned that if I tell colleagues to trust HIBP, I'm priming them to fall for anti-malware protection rackets.

How about the venerable NYT? They must have some great stuff. But then there's also content such as TFA, in which a graphics director and a writer from the opinion desk deliver bons mots such as: "None of us really has a choice to participate in tracking or not — the system just serves up location data, usually without us noticing. […] When we participate in this system, we're tacitly endorsing it." On a page that attempts to contact DoubleClick and Google Analytics. (Which I did notice, and, unlike the NYT, chose not to participate in.)

When someone asks for advice or I noticed them make a security boo-boo, I can teach them the bare minimum (basic stuff such as how to use unique credentials for every site and service). But when it comes to doing my part in spreading the word about being more proactive than the bare minimum … halp!

Edit: Credit where credit is due: I just finished reading TFA and discovered the section "What about The Times?", which has links to "This Article Is Spying on You" [0] and "How The Times Thinks About Privacy" [1].

[0] https://www.nytimes.com/2019/09/18/opinion/data-privacy-trac...

[1] https://www.nytimes.com/2019/04/10/opinion/sulzberger-new-yo...

← PreviousPage 3 of 5Next →