hytheaway

(.github.io)

journal


the desk (may 2025)

it's been a while since i had a photo in the journal, so i'll make up for that here.


what do you notice about this photo?

okay, probably not much. it's a laptop connected to a couple monitors. a keyboard, a mouse, some headphones, books, nothing worth a second thought, and you'd be entirely right. that's an m2 max macbook pro, perfectly capable of driving two montiors and plenty of peripherals through a single thunderbolt port. no big deal.


okay, well, what do you notice about this photo?

well, uh. as long as you know that's a base m2 macbook air, that kind of changes things. base m2 macbooks only support one external display, but i've got two here. i promise, there are no tricks here, and in fact, it gets even more baffling.

for one, yes, there is only one thunderbolt cable being used to attach both of these laptops, and i'm only using one dock, which is the caldigit ts3 plus. the ts3 plus is great, but it doesn't support video through the thunderbolt passthrough/daisy-chain. when you connect one thunderbolt cable to this dock, you can output to one external monitor. but, hey, thunderbolt 3 handles up to 40gbps. that's a lot of bandwidth, may i found some clever way to utilize that extra bandwidth. after all, while my displays are both 1440p and 144hz, that's well under the bandwidth offered by thunderbolt 3, even without display stream compression.

that's where the second point comes in: i'm not even really using thunderbolt 3. no, that cable connected to both laptops isn't a "full" thunderbolt 3 cable. it's a 20gbps thunderbolt cable, all because i use a thunderbolt switcher. and you know what? i'm going to go on a tangent here about this switcher because i've got a lot to say. i hate having a bunch of cables. for my laptops, it makes it really annoying to dock and un-dock. for desktops, i always lose track of which cables i do and don't have connected, and into which ports (i've had kvm switches that refuse to work with anything besides specifically usb 3.1, including 3.2), and it wouldn't be too big of a deal if i hadn't just rebuilt a new pc in this ultra small form factor that can so easily be thrown in my backpack. so, if i ever want to throw this tiny (amazing) pc in my backpack and take it somewhere i'm working, or to a lan party, or whatever, then it's a total pain to be plugging and unplugging a bunch of cables over and over again. as a result, i kinda love thunderbolt, since i can just hook it up to a dock that i'd keep anywhere i'm working and everything just pops up perfectly, and i made sure that thunderbolt compatibility was nearly at the top of my list when i was building this pc. but, it'd be stupid to have two docks at the same desk in my apartment, one for my laptop and one for my desktop, so instead i wanted a thunderbolt switcher.
so why do nearly none exist on the market? i mean, yeah i know why, no one has a need for it. the whole thing about thunderbolt docks is that they made expansion easy with one cable, which is only valuable if you're moving the computer around a lot, and basically no one is moving their desktop around. but, still, i would've paid so much more money for a much better switcher than this one, but this is the only i was able to find on the market at all, let alone the only one within a reasonable price point.
okay, so, why am i saying all this? because with only one product available, you have to be okay with all its features and dis-features. and what dis-features does that include? well, sure, 20gbps, but more importantly, this switcher doesn't support more than one external monitor. so, even if i did have a dock that was able to drive two external displays over one thunderbolt cable, this switch would've prevented that. and furthermore, i took the stupid switcher apart and the whole reason it's limited to 20gbps is because they cheaped out and only used one thunderbolt controller, so the entire 40gbps bandwidth just gets chopped in half between the two inputs. or outputs. or however you want to think about it.

for the well informed, this all seems like i'm leading up to displaylink adapters. and, yes, you're right, that's exactly how i got around this. my signal flow is computer -> 20gbps switcher -> caldigit ts3 dock, which i use the dedicated output for my main display and a displaylink adapter for my secondary display. displaylink is effectively video over usb, so it requires drivers (and permissions on macos) and is limited by usb speeds, topping out at 60hz. honestly, that doesn't bother me. sure, once you've tried high refresh rate, it's hard to go back to sub 90hz speeds, but i usually only have spreadsheets, references, papers, discord, or zoom on that display anyway. even if i have a youtube video playing, 60hz is more than enough.

i've kind of fallen in love with this displaylink adapter, by the way, because it actually overcomes a major hole i didn't realize i had dug myself into. when i was building my new pc, i (apparently stupidly?) assumed that a thunderbolt 4 interface would probably natively support at least two displays at once. i mean, thunderbolt 4 didn't increase the bandwidth from thunderbolt 3, but does overhaul the data transfer symmetry, topology, tunnelling of video, and (in later versions) even does dictate multiple displayport interfaces over the protocol. however, what i didn't consider was that, uh, no one cares about the actual thunderbolt spec and they kinda just implement whatever the hell they want to and leave out whatever they don't want to support. this wasn't an issue until i realized that my brilliant and cute little mini itx motherboard with thunderbolt 4 only supports one display over the connector.
now, this isn't as big of a deal with my desktop, i could always just plug in just one displayport cable and get the full 144hz, but after all this work, after all this planning and trial and error, all to maintain a one-cable solution for my pc, it was unreasonably painful to have gone through all that just to end up with my perfect solution, minus one cable. now, obviously, i also have the power cable plugged in. that computer draws like 450 watts, there's no way the 87 watt charger in the caldigit would suffice. and, reasonably, i also have the antenna plugged in, because the bluetooth doesn't work without it (maybe this is because of the metal case, but whatever) but i could always replace the antenna with a more compact one in the future.


so, where does this leave me? like many people my age in similar industries, i spend a large, large amount of my time sitting at this desk, in this chair in my small, ~400 sqft apartment in midtown manhattan. don't get me wrong, i love this apartment. but what it means is that i try to be intentional with this desk. if there's something on this desk that i don't use, it leaves the desk - there's just not enough space. while we're on this topic, and on the topic of noticing things, one of the things you may have picked up on is my keyboard. going back through this journal, you might see some other keyboards, even one i talked about when i was particularly bored.

i have a lot of keyboards. or, i used to. i've gotten rid of a number of them, and i'm sure i could get rid of more. but i do so much (if not all) of my work on a computer, and the way that i interface with that computer is really important to me. in undergrad, i was often graded on how well i could use a computer without touching the mouse (especially in terms of pro tools), and once you get kind of used to it, that level of fluidity feels great. the reality is, your brain often moves faster than your body, and every time i had to move my hand over to a mouse, just to move the mouse, just to scroll, to click, to double click, all to open a file, it began to feel more and more like a waste. the further reality is, it's not a waste to most people. if you don't spend so much time at a computer or with a specific program, then you'd probably lose more time trying to learn a bunch of keyboard shortcuts. so, my only thoughts on it are, some things are important to some people because of their life, and some things aren't as important to others because of their life. for me, having a good keyboard that feels second-nature, responds well, and doesn't fatigue me is important.
i've really gravitated towards smaller keyboards. the difficulty of using function layers isn't nearly great enough to outweight the wasted space of using a bigger keyboard to me. recently, i've been using the hhkb professional hybrid and it has basically ruined all other keyboards for me. it's small, it can connect wirelessly to up to 4 devices, it doesn't sound like a shotgun to anyone around you, it feels great, i could go on and on. and i will. they keycap profile feels great, i love the position of the control and delete keys, and it feels incredibly durable. not durable in the way that it'll stand up to a sledgehammer the way that a solid block of aluminum would, but durable in the way that if i threw this in my backpack, i wouldn't think twice about it. i wouldn't be worried about the keycaps falling off or chipping or some anodized finish getting scratches from my hard drive or anything stupid like this. it feels durable because it feels like i can be reasonable with this thing, and it'll be reasonable back to me.

enough about the keyboard. onto the mouse. i told you, everything is intentional. i haven't gone through as many mice as i have keyboards, partially because i don't interface with it as frequently, partially because i don't know what makes a good mouse, and partially because i feel like i got super lucky with one of my first picks. the mouse on my desk 99% of the time is the logitech g305. it feels perfectly right in my hand (not too long or wide), it has all the buttons i need (i love forward and back buttons, but i really wouldn't mind having one more button somewhere on it), it has adjustable dpi (i move my mouse with my wrist instead of my arm, so my mouse sensitivity is usually quite high), and, probably most importantly, it's just about the perfect weight.
that last point has become super critical with me lately. over the past few years, i tried out a couple different mice that were supposed to be (and in many ways, were) better than this one: the logitech g pro wireless, and the logitech mx master 3s. i mean, both were certainly priced considerably higher than the g305, and it made me very anxious buying them just for an experiment, but i was curious on what i was missing out on. besides, i used to daily the original mx master something like 6 or 7 years ago until mine broke, and i remember liking it a lot. surprisingly, i've never realized i'd be so subject to the whole "goldilocks" dilemma. the g pro wireless was just way, way too light. it felt like a cheap toy, and it was so light that i actually felt like i wasn't able to predict where my cursor was going to land by the end of it. i know this is normal, and that after an adjustment period, it'd probably be perfect, but besides it's weight, what other advantage did i get for the extra $60 i paid? it's slightly longer in the hand, which i guess feels alright, and it has rgb on it, which doesn't matter at all to me because it's positioned directly under my palm, so whatever. the final nail in the coffin, and by far the biggest gripe i had with it, was that the battery life was abysmal. i think maybe the g305 has spoiled me, but i felt like i was charging the g pro wireless nearly every single time i sat down at my desk. obviously, being a "super-light" gaming mouse, the battery must be tiny. but i need to get work done when i need to get work done, so that resulted in me keeping the charger draped across my desk, and if it's going to be draped across my desk and always at risk of the battery dying, then why would i ever unplug it, and if i would never unplug it then why the hell did i spend $100 on a wireless mouse. so, that one didn't last very long.
the mx master 3s is just too heavy. that's really it, i don't have nearly as many gripes with it. sure, there are a few, but they're no big deal. and realistically, i know the mx master 3s isn't that much heavier than the g305, but the physical size is so much larger that it feels like i'm holding a rock.
there is another mouse i use pretty frequently, at least in comparison to any other mouse, and that's the kensington expert trackball. probably one of the best mice out there.

anyway, the other stuff on my desk is usually just a water bottle, headphones (usually hd650's, but often also 1000xm4's and corsair void), books/papers i'm reading or referencing, guitar picks, a light bar, a screw tray (for whenever i take things apart on my desk), and my stream deck, which handles all my macros. i try not to rely on macros over shortcuts, because i'm not always at my desk, but for things like homeassistant, media playback controls, and other operations that are completely doable with a macropad but just annoying (adjusting per program volume in particular), i keep those close on the stream deck. now, obviously, the stream deck should not be confused with the steam deck, which i also have at my desk. honestly, this one is the most superfluous item on my desk, it doesn't have to be here, but i like how it looks, it makes it convenient for taking short 10 or 15 minute breaks from whatever project i'm completing, and having it on the same desk as my laptop means that when i grab my laptop to pack away, i can very easily also grab the steam deck and not think too much about it.


that's it. that's my desk. i won't live in this apartment forever, so whenever i want to look back and find out how i had this whole thing set up, then i'll take a look here, and maybe write a new update for posterity.

3d printed tower that holds my usb hubs, mini network switch, ts3, thunderbolt switcher, and scarlett solo.

my pc case is the velkase velka 3; steam deck and ps5 controller for scale.

desk from above.

ai (aug 2024)

(for variety's sake, we're on the right side for this one.)

ai is slop. everything done with ai is slop. this isn't an opinion.

ai, which has become shorthand for gen ai, which has become shorthand for generative ai, which has become shorthand for generative artificial intelligence is what happens when you take a freshly created homunculus and feed it the world's knowledge without discretion. chat gpt is what happens when you start asking that homunculus questions.

did you catch that? it was subtle. blink and ya miss it.
here, i'll slow it down for you:

without discretion.

an inherent property of using ai like this is that it will create exclusively slop.
sure, you can add all your fancy little weights and prompts, pathetically begging your machine not to tell people to eat elmer's glue.
but if you're at that point, you might as well start over. another homunculus down the drain.
and if you start over, are you going to feed it the same material? of course not. that would just result in the same homunulus you just discarded.
this time, you'll be smart. you'll be crafty. you'll be mama's little genius.

no, this time, you'll only feed it some of the world's knowledge.
maybe it'll be a finance homunculus. or a material sciences homunculus.
your smrik breaks into a laugh as your belly fills with content. you've fixed it. you've solved the problem, and you did it before anyone else could think of it.
as you lean back in your pleather armchair, sipping watered-down franzia from a snifter, you begin to rent out these homunuli as consultants to financial institutions and chemical research labs. they've been trained with more relavent knowledge than any human could contain.

they'll be better at any job because they'll know more about any single job. and, you remind yourself, the only thing that makes someone good at their job is how much they know about it.
but, as you close your eyes and let the wave of self-satisfaction wash over you, a voice nags from the corners of your mind. ...crit..al th...ing... you can't quite seem to shake it. ...critic.. ..inking... what is that nagging? you couldn't have missed something.
.
..
...
...critical... ...thinking!!!
you shoot awake, eyes buldging half out of their own sockets as your hand races and fumbles its way into your pocket to reach for your phone, spilling napa valley's finest all along the sleeve of your child-labor tweed jacket and hulk funko pop.
it's too late.
the homunculi you sold to these businesses are utterly incapable of doing the jobs that these companies leased them for.
with a shaky hand, you draw back the curtain and look outside. the mob has already assembled, pitchforks and medieval torches.


the problem with imbuing a homunculus with all the world's knowledge, even on a simple topic, is that it will only ever know the work that has been done before.
it will only spit out the problems that have already been solved. you can't teach a homunculus knowledge that doesn't exist yet.
but, this homunculus knows everything, right? so, what happens when you ask it a question that doesn't have an answer?
it'll hallucinate. it'll believe that, if it's been asked a question, then there must be some sort of answer.
and, can you blame it? it has never known of a world where "i don't know" is an acceptable response.

these homunculi lack the ability to take their vast amount of knowledge and apply them to solve a unique problem.
in reality, that's what critical thinking is.
at most, they've become repositories of knowledge. ask them any question, and they'll give you an answer (correct or not).
ask them to do a task, and the best they can do is find a source of someone else who has done that task. they're knowledgable, not intelligent.
unambiguously, the delineation between "knowledge" and "intelligence" is application.

these homunculi were never going to be able to replace the workers. every job, no matter how menial, requires a level of critical thinking that far exceeds the operating potential of any homunculus. even if you, say, gave the homunculus mechanized legs and arms, preprogrammed to execute a basic series of tasks after being called by the homunculus, the homunculus' tendency to hallucinate when encountering unexpected circumstances necessitates a non-homunculus worker to validate every task or batch of tasks they complete, if not having to be peering over their shoulder constantly.

crucially, this will not improve over time.
the homunculus will not wake up one day, suddenly equipped with refined critical thinking skills - it just won't happen. the homunculus is not a critical-thinking machine. there is no latent potential to unlock. it is absent all mental faculites besides database-like storage and democratically elected responses based on how frequently an answer appears in the knowledge-base given to them.
further, sure, you could tell it to never recite an answer that is negative in nature (re:elmer's glue), and it will obey that order. that is, until someone else gives it the exact opposite order. the homunculus knows no master, and it won't obey you more just because you created it.

this might beg the question, why bother using a homunculus if it is prone to so many shortcomings? why not a more intelligent being? a loyal dog? a crow?
because these shortcomings are by design.

for what is a homunculus, but a vessel to project upon?

--- as of writing, we're encountering a new problem with what we call ai.
it has been unleased to the world, and the content that it generates is now being added to databases, the same databases that it pulls from to generate new content.
it's goal is to generate facsimiles of human-born content, generally in an inscrutible way. often, so inscrutible that the arms race between ai content generation and ai-generated content checkers is constantly in flux. if we, as discerning people, are having great difficulty creating tools that can reliably tell what content is ai generated, how can we expect a complex algorithim designed to mimic human approaches to?
we can't, and more importantly, we don't. it's easy money not to.

"who cares, right?",
"we'll leave that problem to someone else",
"it doesn't affect us",
so on and so forth.

yeah, dude, that's a cool perspective to take and all, but your pet robot is beginning to choke on its own vomit.

the immersion question (jul 2024)

by far, the question that i dislike the most within the world of immersive audio.

it's been a moment since i last posted. unfortunately, most of my time has been dedicated to projects that i am unable to talk about here, either for the time being or forever. however, there is always room to talk about one of the most infuriatingly meaningful and meaningless debates within the vast world of immersive audio:


what makes audio immersive?


on the surface, it's a pretty obvious question. i mean, clearly we need to work out what it means for something to be "immersive" before we can go around calling whatever we produce "immersive".

okay, well, how about this: "audio is immersive if it takes advantage of the unique properties offered by immersive audio systems."
yeah, this is definitely a good start, but still only getting about a little bit of the way there. but, what makes an audio system immersive? having a bunch of speakers? then, what about binaural audio? or VR audio? then, maybe an audio system is immersive if the audio sounds like it's coming from all around you. but then, that's not the system's doing, that's the content. and, if the quality of immersive is exclusively content-based, then how do we distinguish systems that are built with immersive content in mind?

so, the answer must be either somewhere in-between "audio is immersive if the system is immersive" and "audio is immersive if it sounds immersive", or must be encompassing of both.

the topic is still very much up for debate (or more accurately, interpretation), but people seem to have generally landed on a definition that includes both, delving specifically into the psychoacoustic attributes of immersion. in some technical way, an audio system is only immersive if it's capable of delivering an immersive experience, which could be done with as few as one speaker/sound source. but, that's at odds with the commerical market, which has specifically created a subset of "immersive audio" products that center around multichannel configurations and their folddowns to binaural.

ultimately, this ambiguity will be either difficult or entirely impossible to resolve, and it's not so much a matter of narrow or broad interpretations. what consitutues an "immersive" experience to one person may not evoke the same reaction to another. one person might find a stereo mix utterly all-encompassing, while another might find it flat and unconvincing. likewise, two different people, both perfectly in the sweet spot, may not agree on how compelling an atmos mix is.

this makes it sound like i'm on the side of those who believe that the content is what makes audio immersive, regardless of its associated technical capabilities. in that case, it would be easy to say that the purpose of a system is what it does, and to end the conversation there.

however, i cannot entirely agree with that perspective, because a full adoption of that mentality inherently either states that for each person in the world, there is at least one piece of audio content that they would describe as "immersive" (to the same or nearly the same extent as if it were played back on an immersive system), even if it was played back on a system not intended for immersive audio experiences, or states that there are some people with full faculites intact who cannot experience audio as immersive. if a person requires a purpose-built immersive audio system in order to experience audio as immersive, then that demonstrates that content alone is not sufficient in creating an immersive experience.

on the other hand, it can be an equally poor perspective to believe that what makes audio immersive is the system its played back on. i don't imagine the reasons for this are all too difficult to determine, but this perspective's simplest and most compelling failing is that, no matter how many speakers or orders you have, a bad multichannel/binaural mix will never be immersive.
remember that the purpose of a system is what it does. therefore, an immersive audio system that fails to always produce immersive audio is not an immersive audio system. it is, at best, a "sometimes immersive audio system".

so, why do i hate this question so much?
cause it sucks, dude.

there's no concrete answer to it, and that's not a bad thing by itself. what makes that a bad thing is that, because there's no concrete answer to it (and in my opinion, there never will), people love to spend all their time in a circle with their hands on their chins postulating on the meaning on immersion and life and humanity and its applicability to audio, instead of going out there and actually putting those good brains to proper use. these people (a vocal minority, i will add) thrive on the very nature of this unanswerable question, because it means that they get to push their thoughts forward, hear themselves speak, correct other people, trample over unrelated conversations, but ultimately take no responsibility for any of it under the pretense of "well, it is an unanswerable question" with a smirk on their face. the question of "what makes audio immersive" becomes the ulimate game of "ummm, actually" in a desperate attempt to prove themselves more knowledgable about a wholly self-constructed classification, topic, and argument, and it's completely disengaging to anyone trying to get a feel for this field.

yes, it's an important question, but what good is that if no one actually cares about the answer? has the lack of an answer stopped literally anyone from progressing research in the field? has the lack of an answer prevented any company from attempting to establish a vertical for themselves in this brand new form of content creation and consumption? in fact, the lack of an answer doesn't even prevent any single content creator from engaging in the field. but these petulant, pedantic, self-aggrandizing members of a mutual admiration society certainly do prevent greener individuals from creating the very content that this field is so desperately in need of.

honestly, these people are a small, small, small subset of those discussing the question of immersion. there have been some amazing papers published that do make a genuine effort to investigate the qualities of immersion, and how best to apply those finding to content created for immersive experiences. i've long felt that the question is important but the debate isn't. the field of immersive audio has reached a point where the technology is having a difficult time developing any further without a sizeable amount of content being produced for it, and while most of the reason for that (in my opinion) has to do with the significant workflow changes associated with creating immersive content, we should be focusing on minimizing any and every other deterrant for newcomers. no one wants to pour their valuable time and energy into a field they are constantly being told they don't understand.

you don't need to educate every person you meet. people won't remember you for being smart, but they will remember you for being overbearing. so many people involved in this field are smart in so many ways. distinguish yourself by being kind. because, in a world where you can be anything, why would you choose to be anything but kind?

a return to form (may 2024)

and it is!
i mean, over function, for sure.

anyone purchasing a ps2 keyboard in the year 2024 probably knows that:
- they're incompatible with modern computers
- they don't support hotswap, so anytime it's plugged in or unplugged, the computer has to be restarted
- usb compatibility relies on third party dongles that are often hit or miss
- the price of good quality ps2 keyboards is rising

but when i was walking back to my apartment the other day, i stopped by a hardware shop to search for any fine point soldering tips (long shot, i know, but it was on the way), when i met face to face with this beauty, still in the box. no surprise there, though. these are still manufactured, after all.

but, hey, how often do you get to try out a rubber dome keyboard that looks straight out of the 80s?


and odhrán seems to like it too.

fpgbc (apr 2024)

akin to that one flight i took that one time.
overwhelmingly exciting for what's ahead, but a bit of a rocky start.

reproducing "retro" consoles with fpgas at incredibly accessible prices is an amazing future that i hope to see.
i can't imagine the complexity associated with reverse engineering a console at all, though.
analogue has been doing this for years, but their business model has devolved to scarcity tactics, their prices are insane (with reportedly absurd shipping prices) and their marketing revolves around the idea that their products are "no emulation". my reach is limited, but let me communicate this clearly:

hardware emulation is emulation.

they even clarify that their products use no software emulation in the small print of their website - they're aware that this claim is unambiguously false.

hardware emulation does suffer from inaccuracy issues, the exact same way that software emulation does. inaccurate (albeit infrequent) cycle speeds, rtc, write/read compatibility, color reproduction, etc. - these all contribute to a less than authentic experience, and the analogue pocket is no different. and yet, it seems like all i hear about the analogue products is from people frothing at the mouth saying "but it's not emulation so it's better!!1!", and analogue directly benefits from this misconception.

so, i'm not about to drop $220 (+$60 shipping!!!) on a product i don't need, to support a company whose business practices i believe are immoral, and whose marketing tactics so carefully dance the line of deception that it's remarkable all they've been required to disclaim is in small print on their website.

so. the funnyplaying fpgbc.
imagine if the analogue pocket was maybe 10% worse at it's aimed goal, about 120% less expensive, and about 50% more fun.
at like $85 all in, with free shipping, it's not just one of the easiest diy kits for the gameboy family, but you also don't need to have a gameboy or gbc to mod.
this entire device is kind of sold as if it's just another mod kit for gbc, but no - it's entirely its own console.

that's not to say it's without it's issues.
the most glaring one is that it doesn't seem to work with most reliable flash carts, including krikzz's everdrive-gb, and the ez-flash junior (which is the one i have).
beyond that, there are a few options for the display, including 4x integer scaling and a non-integer scaled view that fills the whole screen, which obviously will include shimmering, but honestly, i don't care about that. it looks amazing filling up the whole screen. so what's the problem? well, for each of the sizing options, there's a version that includes a color profile that attempts to emulate what a gameboy color looks like.
it's honestly not worth complaining about - just don't use it. turn it off, and don't turn it back on. it doesn't look good, i don't feel like it's particularly accurate, and i await a day when a firmware update patches it properly.

i've had the fpgbc in my backpack with my copy of pokémon crystal since i put it together, and i've loved it. maybe my cartridge is borked, but the rtc doesn't work. i haven't tried it with my actual gameboy color (see dec2023 entry), so while i cannot fathom why it would be an issue with the fpgbc, i still don't care to rule it out as a possibility.
i do believe that the best game console is the one you have with you, the same way the best flashlight is the one you have with you, the best knife is the one you have with you, etc. etc., so it should come as no surprise that this has quickly become my more favorite game console recently, even with its flashcart shortcomings.

love this little thing.



gameboy color mod no. 1 (dec 2023)

can you spot the difference between these two?


i mean like obviously yes you can they look incredibly different but that's besides the point.


more like, what would you expect to be different between these two?


i'll tell ya, not much.


but it's cool enough.