bitshift's recent activity
-
Comment on Those who journal, how do you do so authentically? (How to stop “self-editing” or “censoring” yourself and your thoughts?) in ~health.mental
-
Comment on Extraverted introverts, cautious risk-takers, and selfless narcissists: A demonstration of why you can’t trust data collected on MTurk in ~science
bitshift Link ParentThat's wild that there's active subversion going on, as in people putting in significant effort to cheat the system, and to cheat more effectively. I shouldn't be surprised, but I had assumed...That's wild that there's active subversion going on, as in people putting in significant effort to cheat the system, and to cheat more effectively. I shouldn't be surprised, but I had assumed cheating was more low effort, such as mashing keyboards.
You said you built apps on the platform? Were you (or your clients) in the position of "employer", then? If so, I'm curious what your approach was for combatting cheating.
-
Comment on Making tough decisions: what’s your go-to approach? in ~talk
bitshift LinkI have a particular method for sleeping on tough decisions: Get some paper and write down the factors that matter to you. These can be concrete things, like "money" or "time commitment". Or they...I have a particular method for sleeping on tough decisions:
- Get some paper and write down the factors that matter to you. These can be concrete things, like "money" or "time commitment". Or they can be abstract: "satisfying my curiosity" or "gut feel". Whatever factors you value when making your decision.
- Write down your options. Should I take up knitting, ultimate frisbee, or lion taming? Tough choice.
- Use your factors to score each of your options. For example, I might give knitting 5 stars on time commitment (I can do it on the bus), 1 star on money (I have expensive taste in yarn), and 3 stars on gut feel (I kinda want to give it a try). Score the remaining options the same way.
- Add up the stars. Knitting is 5 + 1 + 3 = 9 stars. Ultimate frisbee got 3 + 5 + 4 = 12 stars. Lion taming: 2 + 4 + 1 = 7 stars.
- Put the paper away and sleep on it. The next day, score your options again. Try not to look at yesterday's answers! This time, knitting is 10 stars, frisbee is 12, and lion taming is 6.
- Do this several times. If you're in a hurry to get more data, you can do it 2-3 times per day, spread throughout the day so you have a fresh mind. You might notice scores trending in certain directions: e.g., frisbee is getting fewer stars each day you think about it.
- As soon as you have enough data, get out a scientific calculator and average each option's score. You'll want to use an exponential moving average with a base of 1.618, the golden mean— just messing with ya! Don't use the numbers. They were just a device to force you to think about the problem carefully, from all different angles. And by this point, you should already know in your heart you were meant to tame lions.
-
Comment on What programming/technical projects have you been working on? in ~comp
bitshift LinkCasually working my way through Crafting Interpreters, though I'm substituting a language of my own design in place of Lox. It's a macro language for code generation. Think assembly language...Casually working my way through Crafting Interpreters, though I'm substituting a language of my own design in place of Lox. It's a macro language for code generation. Think assembly language macros: including files in the middle of other files, deduplicating repetitive code, evaluating template strings with numerical expressions, etc.
As part of designing that macro language, I've been studying languages that solve similar problems, such as actual assembly languages. My limited experience with assembler macros has been that simple things are easy, while everything else is difficult/impossible. Contrast with Tcl, which offers extreme flexibility — but don't goof up your syntax, because you'll have a heck of a time finding the issue. For my own language, I'm trying to strike a pleasant balance between these extremes.
I'm pretty far from my ultimate goal (the side project for which I'm inventing this macro language), but I'm enjoying the journey, and that's 90% of the satisfaction I'm hoping for. But I do hope to reach the destination someday! I'll have to fight perfectionist urges to get there, though. The pragmatic Rust mantra of "Keep calm and call clone" is extremely relevant here, as is the general attitude of not overcomplicating things. For example, do I really need parser errors to have line numbers? Probably not. But stuff like that is tempting.
-
Comment on What's something you've been mulling over recently? in ~talk
bitshift Link ParentSorry to hear about your husband's health issues! Besides the logistical stuff like time and money, it's never fun to see someone you love suffer. It sounds like your sister's from a more...Sorry to hear about your husband's health issues! Besides the logistical stuff like time and money, it's never fun to see someone you love suffer.
It sounds like your sister's from a more conservative branch of Christianity? FWIW as a more liberal Protestant, I hope she's incorrect about some of God's aspects. I don't have satisfying answers, but the unsatisfying summary is I hope there is much more goodness and mercy out there than that, somehow.
I like your use of the word "Lovecraftian". Sometimes I ponder how wrong I might be about the universe, e.g., what if we're in a simulation, and whoever's in charge is Lovecraftian in the sense of cosmic indifference to our suffering (as opposed to an entity that inflicts harm for harm's sake). After all, I run computational stuff all the time, and I've never lost sleep over whether I've accidentally created suffering! In fact, we give strange looks to people whenever they say stuff like LLMs might have consciousness. If the hypothetical Programmer Upstairs was anything like a human programmer, it would be bad news for us.
To be clear, I don't believe we live in the Matrix — I'm fond of Maciej Cegłowski's criticism of the simulation hypothesis as "a powerful solvent for sanity." But I can't rule it out, which is kind of distressing. Ultimately, I cannot change the universe I'm in, so I try not to worry about it — it's just one of many unanswered questions, alongside such classics as, "Why is there something rather than nothing?"
-
Comment on What programming/technical projects have you been working on? in ~comp
bitshift Link ParentRegarding cryptography, my impression is that they've done their homework but are constrained by poor historical decisions. Off the top of my head, the big one is encrypting data using AES-CTR:...Regarding cryptography, my impression is that they've done their homework but are constrained by poor historical decisions. Off the top of my head, the big one is encrypting data using AES-CTR: this works, but it comes with extra assumptions about the rest of the system behaving well, assumptions that aren't very safe to make. And unfortunately, changing Borg's encryption algorithm is difficult to do in a backward-compatible way.
My anxiety about it is currently at an acceptably low simmer. Though if an attacker was actively modifying my cloud data in an attempt to break Borg, I'd be a lot more worried! Now I'm curious to look into how Kopia does things under the hood.
-
Comment on What programming/technical projects have you been working on? in ~comp
bitshift Link ParentCoincidentally, I've also been revising my backup system — though my particular setup is a couple of Linux machines using Borg to write to rsync.net. Previously, running ad-hoc Borg commands...Coincidentally, I've also been revising my backup system — though my particular setup is a couple of Linux machines using Borg to write to rsync.net. Previously, running ad-hoc Borg commands involved copy/pasting environment variables, but now a brand-new shell script manages everything for me.
How has your experience with Kopia been? Borg has been pretty good for my use cases. The one big limitation I'm aware of is that you need Borg running on the remote end as well, so you can't easily put backups in key-value based cloud storage like S3/R2. (rsync.net has also been very good to me, and they have Borg installed on their servers, but it does limit the cloud providers I can consider.)
-
Comment on Downtime due to sign up spam in ~tech
bitshift Link ParentSpeaking of Hashcash, the papers section on their website is delightful reading, specifically some of the later papers under "Related Work". There are some more recent proof-of-work schemes which...Speaking of Hashcash, the papers section on their website is delightful reading, specifically some of the later papers under "Related Work". There are some more recent proof-of-work schemes which emphasize memory-hard problems — RAM and memory bandwidth being where ASICs start to lose their advantage.
The clever bit is that these schemes can be designed to be memory-easy to verify, yet memory-hard to compute. For example, the client might have to find N different values whose hashes all have the same prefix; the client's best strategy is to generate many more than N hashes and store all of them in memory until the birthday paradox kicks in. But once the N values are found, the server's work is easy: hash those N values and make sure the prefix doesn't change.
-
Comment on Tell me about your weird religious beliefs in ~humanities
bitshift Link ParentThe interesting thing to me about your thought experiment: it's the simulation hypothesis, but with unlimited recursion. Normally you'd expect each nested simulation to be "smaller" than its...The interesting thing to me about your thought experiment: it's the simulation hypothesis, but with unlimited recursion. Normally you'd expect each nested simulation to be "smaller" than its parent in some way: if not in complexity, then in terms of the storage space required by the simulation. Eventually, the recursion must end.
But if you don't need to physically evaluate the Game of Life, you can nest simulations to your heart's content. Imagine a Life pattern that implements a complete Turing machine: one property of that pattern is that it contains every possible universe that can be simulated by Turing machine, and those universes all experience existence regardless of whether you evaluate the pattern. And if your own universe is simulatable, then the pattern literally includes your own universe — not in the sense of "simulating a copy of a subset of your own universe," but in the sense of "these two mathematical objects are the same."
I can't say that I seriously believe it, but it is a lot of fun to think about.
-
Comment on Why does the letter 'S' look like an 'F' in old manuscripts? in ~humanities.languages
bitshift LinkI just remembered that the Greek language has a similar thing with the letter sigma. Wikipedia gives Ὀδυσσεύς (Odysseus) as an example: the first two sigmas look like "σ", but the last sigma in a...I just remembered that the Greek language has a similar thing with the letter sigma. Wikipedia gives Ὀδυσσεύς (Odysseus) as an example: the first two sigmas look like "σ", but the last sigma in a word looks like "ς". And there's also a lunate sigma ("Ϲ", lowercase "ϲ") whose usage feels similar to long s. Wikipedia again: "In modern, edited Greek texts, the lunate sigma typically appears primarily in older typesetting."
Now I'm wondering if it's at all connected to long s, because it's a funny coincidence that both languages would have special rules for the character that produces the "sss" sound.
-
Comment on Interview: Kenta Cho (Japanese indie game developer) in ~games
bitshift Link ParentFrom my point of view, "content" is just an axis along which you can measure. A haiku can be deep, but in a different way than The Little Prince is. And neither one is the "correct" deep! They're...From my point of view, "content" is just an axis along which you can measure. A haiku can be deep, but in a different way than The Little Prince is. And neither one is the "correct" deep! They're just different dimensions.
When you hear "indie game", you might think of something like Cave Story, and if that's what you expect Kenta Cho's games to be, you might be disappointed. What he writes seems to be more of a "poetry of games" — your haiku metaphor is an apt one.
-
Comment on Interview: Kenta Cho (Japanese indie game developer) in ~games
bitshift LinkThis guy is quite prolific. I remember his name from years ago, when I was hunting for games to install on Debian; looks like he has ~14 packages to his name, but evidently that's far from all of...This guy is quite prolific. I remember his name from years ago, when I was hunting for games to install on Debian; looks like he has ~14 packages to his name, but evidently that's far from all of them:
In 2021 he created a total 139 games, which is one hell of a lockdown project.
Indeed. Sounds like he's invested time into his tools to be able to crank these out:
The goal of Crisp Game Lib is to enable the creation of innovative games within a three-hour timeframe.
So I presume most of these games aren't that deep in terms of content. That said, I presume he's learned something about game design in the process. Makes me want to try one of his games now.
-
Comment on The US GOP doesn’t want to punish trans people—it wants to eradicate them in ~lgbt
bitshift Link ParentI don't think it will go the other way (e.g., prediction markets say the Obergefell ruling will stand), but ultimately we don't know yet, and you're not wrong to be concerned when a seriously...I don't think it will go the other way (e.g., prediction markets say the Obergefell ruling will stand), but ultimately we don't know yet, and you're not wrong to be concerned when a seriously weighty matter is at stake. My thoughts are with you, and with everyone else whose rights are threatened.
-
Comment on The US GOP doesn’t want to punish trans people—it wants to eradicate them in ~lgbt
bitshift Link ParentI keep on thinking about these words, though perhaps in a more hopeful way: bullies are changing their targets because (on average) people are gradually becoming more compassionate and accepting...This isn't a new thing. It's just gotten worse since it's no longer politically acceptable to say these things about lesbians and gays
I keep on thinking about these words, though perhaps in a more hopeful way: bullies are changing their targets because (on average) people are gradually becoming more compassionate and accepting toward those unlike themselves.
At least, I hope so. The bullies are still going to be there, and their attacks will be focused on smaller and smaller segments of the population.
-
Comment on Why AI writing is inherently coercive in ~creative
bitshift Link ParentThis is a good summary of where I stand on allowing ChatGPT in interviews. My interviews are open-book: use the tools you would normally use on the job, including AI. But I expect the same quality...I wouldn't outright reject someone just because I suspected ChatGPT may have been used--the content of the answers themselves is absolutely being evaluated.
This is a good summary of where I stand on allowing ChatGPT in interviews. My interviews are open-book: use the tools you would normally use on the job, including AI. But I expect the same quality of code and depth of understanding as I would want from a future teammate.
I'm reminded of the story a while back of the guy who got fired for outsourcing his code to a programmer in another country. IIRC, "his" contributions to the team were reasonably good, and his ability to orchestrate the whole thing was actually kind of impressive. The only reason anyone found out is because he had to mail his security key overseas, and that tripped some alarm. The code wasn't the issue — the deceit was.
(Similarly, if you're a candidate who's using ChatGPT, show your work and don't lie to me.)
-
Comment on ‘Impossible’ to create AI tools like ChatGPT without copyrighted material, OpenAI says in ~tech
bitshift Link ParentThis is a really interesting point. A lot of the dialogue is around creators and their rights: an artist is entitled to their work for life (plus additional years), and OpenAI is violating their...This is a really interesting point. A lot of the dialogue is around creators and their rights: an artist is entitled to their work for life (plus additional years), and OpenAI is violating their rights by using their work without permission. But are those rights good for society as a whole?
(I suspect no. Ideally, we would return to short copyrights, but I don't have much faith that Congress will do that.)
But supposing we did have short copyrights again… Personally, a 20-year-old LLM would still be tremendously useful to me: maybe it wouldn't know the latest slang, or what a "smartphone" is, but it would still know enough of the English language to function as a writing assistant.
-
Comment on The homemade limits of everyday weirdness in ~arts
bitshift LinkReading the essay, I feel like I have a sliver in common with the author, just as I suspect he has a sliver in common with Calder. I resonate with that sense of not just seeing the world...Reading the essay, I feel like I have a sliver in common with the author, just as I suspect he has a sliver in common with Calder. I resonate with that sense of not just seeing the world differently, but… more than that.
I’ve spent a lot of time with that mobile — hours on end — commonly sitting on the floor in the corner of its Turinese museum room, with a laptop. The Italian museum functionaries are cool with this act of communion. They probably think that I’m writing something about the mobile. I’m not, but I’ve written a lot in there, and been delighted with its presence.
That's the paragraph that got me. I recognize that feeling of being mesmerized by an alien mindset, one that doesn't make sense until you consider that the underlying goals are also different. And there's a compulsion. At the extreme, you end up like Calder, and you build an eccentric life for yourself not because you're trying to be an artist, but because that's the only thing you can do. But in a milder form, perhaps you observe someone else's mobile, understand a fraction of what's going on, and become entranced.
-
Comment on A secret tunnel in a NYC synagogue leads to a brawl between police and worshippers in ~humanities
-
Comment on Is fandom.com actually getting worse? in ~tech
bitshift Link ParentI got the sense of an implied judgement (which might not have been your intention!), and I wanted to let you know in case others did as well — as it would distract from some very valid points...I got the sense of an implied judgement (which might not have been your intention!), and I wanted to let you know in case others did as well — as it would distract from some very valid points you're raising. It's a very Kantian perspective, that if everyone used ad blockers the system would break down, so something is inconsistent somewhere. Even if we reach different conclusions about what the fix is, it's still a very worthwhile discussion.
-
Comment on Is fandom.com actually getting worse? in ~tech
bitshift Link Parent"Piracy" strikes me as the wrong word, but I think I understand the sentiment. A decade ago, I didn't really use ad blockers: ads felt less obnoxious back then, but a large part of my reasoning..."Piracy" strikes me as the wrong word, but I think I understand the sentiment.
A decade ago, I didn't really use ad blockers: ads felt less obnoxious back then, but a large part of my reasoning was because I wanted to support websites I liked, and many of those websites got paid per ad impression. I remember trying out an ad blocker at one point, so I was definitely aware of them. I just didn't use them.
Fast forward ten years, and ad blockers are on all of my devices. But I empathize with your views because I used to hold similar ones. I would like to caution you that your words sound slightly self-righteous on the web — but to the extent that your beliefs are genuinely held, I respect that, and I hope others respect that as well.
I have not done this personally, but one possibility is following Beatrix Potter's example: write your journal in a secret code of your own invention (perhaps in conjunction with writing everything in teeny tiny scribbles, for extra security).