Mastodon Feed: Posts

Boosted by baldur@toot.cafe ("Baldur Bjarnason"):
Tattie@eldritch.cafe wrote:

I suspect a fair bit of transphobia is due to deep-seated instinctive anger against anyone "betraying the team". Why would a person do this? They must surely be getting some personal advantage, it's the only explanation! Selfish! Untrustworthy! Against the rules!

In truth, the fact that people can and do "switch teams" is a huge piece of evidence for existence of implicit gender identity, whatever that might be and wherever it sits.

It is not fucking easy to call into question the team you were assigned by society at birth. Especially when you don't understand yourself what's driving you to do so.
5/

andreu@andreubotella.com ("Andreu Botella :verified_enby:") wrote:

@soatok In my case, I'm lucky to not be worried about being replaced personally, but I am increasingly alienated from the tech industry and dissatisfied with it because of AI. And like you, I have no other career options I could fall back to.

brennan@social.lol ("Brennan Kenneth Brown") wrote:

RE: https://infosec.exchange/@catsalad/117340979884602786

What a wonderful list :)

Boosted by brib@bribstodon.xyz ("brib :neofox_floof:​ :Nonbinary:"):
APBBlue@thepit.social ("Pumpkin Spice Amy") wrote:

Wearing my most obscene T-shirt to the store today.

soatok@furry.engineer ("Soatok Dreamseeker") wrote:

There is no instruction book for being you. There is no machine that can be you. No one but you has your relationships with the other people in your life.

For better, or for worse.

soatok@furry.engineer ("Soatok Dreamseeker") wrote:

If you're afraid of failing, why?

My latest blog post was about my own failure: https://soatok.blog/2026/09/12/the-v8-javascript-runtime-undermined-my-constant-time-javascript-library

Failure is natural. Failure isn't anything to fear. Fail, fail, and fail some more. Then study your failures. Maybe you'll find your own way to succeed.

soatok@furry.engineer ("Soatok Dreamseeker") wrote:

If you're still doubting yourself: yeah, that's natural. You still owe it to yourself to try.

The gay furry without an ounce of real talent in his body is telling you this.

soatok@furry.engineer ("Soatok Dreamseeker") wrote:

And the most important thing to keep in mind is this: while my circumstances and relationships are certainly particular to me, there's nothing special about me.

All that bullshit about "networking" you hear all the time? Well, underneath the mountains of nonsense, there is opportunity to make an impact there. Especially for collective action.

You just gotta look past individual companies. Many of my employer's "competitors" are among my friends. This is healthy. Never let your boss control your narrative or relationships.

soatok@furry.engineer ("Soatok Dreamseeker") wrote:

This is all to say, when AI psychosis is being shoved into the limelight on every platform and they're all trying to scare me, I just laugh and then go eat dinner with friends. Fuck that noise.

soatok@furry.engineer ("Soatok Dreamseeker") wrote:

And, suppose my job was on the chopping block next?

There's not much recourse... if you only look at life through a corporate lens of individuals with skills that fit along a neat hierarchy.

But there is strength in community and in diversity. If you cultivate healthy relationships when times are good, you can trust most of them will not kick you when you're down. At the end of the day, that's all you really can trust. Everything in life is a crapshoot beyond the goodwill of your neighbors.

soatok@furry.engineer ("Soatok Dreamseeker") wrote:

If I was worried about people being able to replace me, I wouldn't have started a blog where I regularly try to teach folks about how to [learn to] replace me, now would I?

Regardless of what happens with LLMs in the next few years, your biggest threat is other people in positions of power failing to understand the assignment.

And, well, you can look into the history of labor movements for ideas on what to do next.

brennan@social.lol ("Brennan Kenneth Brown") wrote:

Kindness as First-class Citizen in Programming

https://brennan.day/kindness-as-first-class-citizen-in-programming/

#programming #coding #ruby #webdev #technology

soatok@furry.engineer ("Soatok Dreamseeker") wrote:

"But Soatok, a machine will be able to do your job tomorrow. Aren't you scared? Aren't you excited?"

No. And no.

So what? Lots of people can do my job (especially if you only have a shallow understanding of what "my job" is). Does that automatically make me unemployed?

My job isn't merely "the guy who knows how to find bugs in code", or "someone who knows how to do the cryptopals challenges" or even "guy that can hack into computers".

My job is "helps people fix security problems especially with cryptography." That requires understanding the client and the client's custoners. It means figuring out (to some extent) their fears, biases, blind spots, bad habits, assumptions, and enemies. And then moving the needle somehow.

A list of machine-detected vulnerabilities doesn't actually do that work. It's just an input; inert and ultimately useless without expert care.

soatok@furry.engineer ("Soatok Dreamseeker") wrote:

The people who are pushing the "AI will replace you" talking points online fundamentally do not understand this. And I worry some well-meaning people will miss it too.

A large component of any employment relationship is trust. Your employer trusts that you will make a good-effort attemptt to fulfill your duties and cultivate a working professional relationship with everyone else, you trust that your paycheck will deposit on time with the correct amount.

Trust is not a simple math equation you can reward-hack your way into solving. Trust is personal, sometimes irrational, cultural, and very social. The closest you could probably get involves cellular automata; and that would only help you study networks of independent actors making trust decisions and how those shape interactions rather than solve trust itself.

You cannot trust any machine the same way you trust the people you work with every day.

soatok@furry.engineer ("Soatok Dreamseeker") wrote:

I'm god awful at any skill that correlates with group activities: music, dance, improv, acting, etc. My creative writing leaves a lot to be desired, to say the least. My illustration skills? Mediocre would be a compliment. I know next to nothing about fursuit making or tailoring in general.

The only things I'm decent at are a combination of software, math, technical writing, and code review.

If there were any kernel of truth to the narratives spread by tech bros today, I should probably be panicked about imminently being "replaced". But such a folly would be missing an important point.

soatok@furry.engineer ("Soatok Dreamseeker") wrote:

Perspective is hilarious and makes life interesting.

neatnik@social.lol ("Neatnik :prami:") wrote:

"...what really influences trolling behaviour is the social pleasure derived from knowing that others are annoyed by it. The more negative social impact the troll has, the more their behaviour is reinforced. Happily, this discovery suggests an easy way to deal with trolls: ignore them, rather than giving them the satisfaction of an angry reaction."

https://theconversation.com/dont-feed-the-trolls-really-is-good-advice-heres-the-evidence-63657

jonny@neuromatch.social ("jonny (nonvenomous)") wrote:

it does appear that editing the files through the app notifies the LLM in some way, it reverted some prior test edits, but it does not appear to notice when i edit the file through the shell. this is unconfirmed - i was in the process of dumping its raw event logs when i had to stop becasue it was eating up all my tokens, and i sure as hell am not going to pay for more - the only way i have found to dump the raw event db is to ask the llm to do it, which it does after some coaxing, but everything needs to pass through its context window as tokens then.

Boosted by adele@social.pollux.casa ("Adële 🐁!"):
lproven@vivaldi.net ("Liam Proven") wrote:

This is a modern motherfucking website.

And it’s still fucking perfect.

https://modernmotherfuckingwebsite.dreamstation.systems/

jonny@neuromatch.social ("jonny (nonvenomous)") wrote:

I am trying to invoke the fleet lessons but my logs all say there is some flag that says "fleet learning disabled for this account" so I think it is not live yet. Sad! In the meantime I think it is so great that the ad industry gets to rebrand "a gigantic database of all our personal data" into "swarm intelligence" by just making the format markdown and using an endless series of nervously chattering dictionaries instead of math as the algorithm.

Boosted by EmilyEnough@hachyderm.io ("Emily 🏳️‍🌈🏳️‍⚧️"):
donni ("donni saphire") wrote:

Every night I worry about 258 things, none of which happen, thanks to my preventive worrying

neatnik@social.lol ("Neatnik :prami:") wrote:

Seeing the Unstory stuff unravel today made me realize that far too many bloggers ignore the prime directive of the internet: don't feed the trolls.

Having a blog with a clean design and witty writing doesn't make someone worth reading or engaging. And if they're anonymous, and mainly post ragebait? Do yourself and everyone else a favor by closing the tab and forgetting about them.

Boosted by pixxl:
CptSuperlative@toot.cat ("Captain Superlative") wrote:

Are LLMs capable of causing enormous harm?

Yes, they are stochastic parrots. So if human beings hook LLMs up to systems that are already potential instruments of harm they will cause harm, stochastically.

But what if we train the LLMs to be GOOD?

The harm isn’t the result of LLMs being good or bad. It’s the result of them being stochastic text extrusion programs.

(Fastidious Ignoring)

Is there anything we can do to prevent these harms?!

Yes, don’t hook LLMs up to potential instruments of harm.

I mean, like, but, is there anything ELSE we can do?!

No.

Boosted by pixxl:
Edmonds_Scanner@universeodon.com wrote:

Your daily affirmation:
The weekend is a great time to start an ambitious new project - one that you have no intention of ever finishing. Think big and fail bigger.

Boosted by pixxl:
bd@merveilles.town ("Bad Diode") wrote:

@andybalaam thanks for sharing, I feel every word of this. As someone who has loved computers, loves learning and making cool stuff, I've also been down as of late.

I'm lucky I'm privileged enough that my employer doesn't force me to use it, but it hurts to see the enthusiasm of a large number of colleagues and people I used to respect. I wish programming was not seen only as a purely technical field where only the outcome is valued, for it is not, it's a creative discipline and code itself has aesthetics and is a reflection of their writers.

Regardless, seeing people making stuff, rejecting the status quo that technocracy is trying to impose on us, and sharing their love for the craft always gives me a boost of energy. So thank you again for sharing, hope you manage to feel better soon.

Boosted by ireneista@irenes.space ("Irenes (many)"):
payown@dragonscave.space ("Michael Babcock") wrote:

Here's what "we'll add accessibility later" actually means: never. If it's not part of the plan from day one, it doesn't survive contact with the deadline. I've watched that promise die too many times to believe it anymore.

Boosted by pixxl:
ploum@mamot.fr wrote:

RE: https://neuromatch.social/@jonny/117325137427138728

Just take some time to understand that a lot of the "magic" and "creativity" of chatbots is basically hardcoded. Thousands of humans wrote it.

The AI hype is basically ELIZA 2.0 powered by a nuclear plant.

Boosted by pixxl:
Tock@corteximplant.com wrote:

The day you stop being a music fan is the day you stop listening to new music. Every generation has a style, a message and a voice worth listening to.

But you can't find any of that with AI. AI cannot innovate, it can only melt. Whatever sounds good with AI is more luck than skill, and it absolutely can't be done live. We know it's weak spots. AI is only a sad, syncophantic weapon. Stuck playing the same stolen music and trained data. It cannot dream like you.

If you are a musician, pick up an instrument, fire up that laptop DAW, or warm up your voice and get to work, Suno be damned. Make utter crap if you have to because you cannot get better unless you make those mistakes. You cannot find what you want to say, what your voice is like... You cannot learn what you want to do, if you are overwhelmed with only what not to do. Take baby steps if you have to. But step.

You don't have to beat AI to win. Play on.

#Musicdon #LearnMusic #Art #neverslop

Boosted by pixxl:
catsalad@infosec.exchange ("Cat 🐈🥗 (D.Burch)") wrote:

Stop using slop machines!
I can also give you misinformation and I'm beautiful :3

#NeverSlop

Boosted by pixxl:
jimbob@aus.social ("ahistorical immaterialist") wrote:

Tag yourself