iiinova — where your digital self keeps evolving

Benjamin Borowski 🍁 @typeoneerror.com · 4h

I'm confident that the fire services in Canada are moving towards better safety, funding, and handling of EV fires. My deputy chief is working on standardized response for our region and his guides/analysis of the concerns is great work. I also like the direction "paid on call" is taking in BC. 🇨🇦🚒

140

Benjamin Borowski 🍁 @typeoneerror.com · 5h

daily #playbleton sketch. some kinda 'wave. 🌊

000

Benjamin Borowski 🍁 @typeoneerror.com · 8h

Was gonna tell some folks "please don't use Muse AI", but I've realized that a significant chunk of my ongoing burnout is in the attempt to convince excited people out of their desire for Progress. The Q for me becomes "how do I still feel the excitement while maintaining a critical lens?" Painful.

030

Reposted by Benjamin Borowski 🍁

Zach Weinersmith @zachweinersmith.bsky.social · 12h

Addendum 2: I'm an artist making art who has an arts degree! I care about this stuff! But sometimes I feel like I'm in a ship that's on fire and saying "the ship is on fire" and people keep saying "Sir, people have always swum."

21219

Reposted by Benjamin Borowski 🍁

Citizen Platano 🇵🇷 @daniloc.xyz · 30/09/2026

can we talk about how my diploma mill film degree ended up being a massive fucking boon right as the agent age kicks off

2433

Reposted by Benjamin Borowski 🍁

Kai 🌟 @kaidav.is · 15/06/2026

the least realistic thing about cyberpunk novels is the protagonists weren't CONSTANTLY complaining about running their own rigs vs the off-the-shelf 'Ono-Sendai Cyberspace II' or spending chapters debugging and rewriting their harness/deck software.

0214

Benjamin Borowski 🍁 @typeoneerror.com · 30/09/2026

i don't suppose harnesses can be done by definition, both by the possibility that tool calls aren't guaranteed by context and predictably at some point you're going to change models. curious what your testing harness knows.

220

Benjamin Borowski 🍁 @typeoneerror.com · 29/09/2026

Other fun facts about Subvert: we just met recently to vote for our new Board of Directors (I put forward myself as a long-shot candidate) and voted on measures such as how Subvert's "worker-members" are hired. Tiny company. I'd love to become a worker-members once more funds are raised.

000

Reposted by Benjamin Borowski 🍁

tef @tef.bsky.social · 29/09/2026

writing a rationalist fanfic about 'who framed roger rabbit?' but it's a series of spreadsheets and charts explaining the economic benefits of a freeway system and justifying the wholesale destruction of toontown

434960

Benjamin Borowski 🍁 @typeoneerror.com · 29/09/2026

no doubt! for all its weirdness, thr place i was a swe where it worked exquisitely was a sociocracy. role-holders set the stage for work rather than a hierarchy. i managed to get myself a "Jira Specialist" role 😁

010

Benjamin Borowski 🍁 @typeoneerror.com · 29/09/2026

unapologetically enjoy a well maintained jira

140

Benjamin Borowski 🍁 @typeoneerror.com · 25/09/2026

Finally a good non-AI Notion update today. You can now control individual property access. Set access to "No access" by default to hide from everyone, then add Group overrides to extend access to specific people. Also nice: preview as!

Screenshot of a Notion "Property access" dialog for the "Readiness" property.

Left panel: Under "Default access" ("Applies to all people, agents, and connections"), the setting is "No access." Under "Exceptions" ("Give specific individuals different access than the default"), "Everyone with full access" has "Full access," and "ADMIN" has "Can view property & values." Below is an option to "Add a person, group, or person property." The footer shows a "Learn more" link and "Clear" and "Save" buttons.

Right panel: "Preview as" is set to "Benjamin Error," showing the result "Can't see property: Doesn't have access to this database."

020

Benjamin Borowski 🍁 @typeoneerror.com · 25/09/2026

I wasn't a big pc games guy as we had a mac ppc, but I found a few.

burn cycle
day of the tentacle
journeyman project: buried in time
super metroid
landstalker
dune ii
xenogears
deus ex
final fantasy vii

010

Benjamin Borowski 🍁 @typeoneerror.com · 25/09/2026

Nice selection. Ever play Dragon Force on Sega Saturn? Always felt it had good Suikoden vibes. I'd love to make a game like that when I retire.

110

Benjamin Borowski 🍁 @typeoneerror.com · 24/09/2026

nailed the drums on last night's sketch—might be one i take to arrangement #playbleton

010

Benjamin Borowski 🍁 @typeoneerror.com · 24/09/2026

sounds about right. a vat of chowder is a safe bet.

010

Benjamin Borowski 🍁 @typeoneerror.com · 24/09/2026

Stuff like: Notion's new Custom Blocks feature (part of its Workers/Dev platform) defaults to creating Vite apps. A lot of early AI prototype stuff followed by buildchain optimization? Who knows; everyone's building the same product with the same tools.

000

Benjamin Borowski 🍁 @typeoneerror.com · 24/09/2026

Top pdx-carpet-shoes photo of all time. What an adventure!

110

Reposted by Benjamin Borowski 🍁

vortex_egg @vortexegg.com · 24/09/2026

The credulity of everybody thinking this is some kind of autonomous intelligence is off the charts. The rhetorical displacement of agency as an accountability sink is the greatest invention of the 21st-century.

27019

Benjamin Borowski 🍁 @typeoneerror.com · 24/09/2026

Fully agree. I've been on a 🎢 coaster: ititial "good" LLMs made me worried, but revenue has increased 20% in the past year and we don't do AI. Good LLMs create business chaos, and I am a chaos enjoyer. Happy to talk agentic with clients, but the "expert" work is always that: talking. With humans.

082

Benjamin Borowski 🍁 @typeoneerror.com · 23/09/2026

couple days and two loads of piss-stained laundry was enough to cure me of this need. 😂 i've realized i am not a dog or a cat person. i was a Mochi person. man, i miss her still.

040

Benjamin Borowski 🍁 @typeoneerror.com · 23/09/2026

hmm, john maus, ariel pink, ruan power, kate nv, enjoy, cowgirl clue, jerry paper, dat politics... kata by kate nv is an obsession tho

youtu.be

KATE NV — KATA

YouTube video by Kate NV

000

Reposted by Benjamin Borowski 🍁

vortex_egg @vortexegg.com · 23/09/2026

The post-information social condition is revealed by the age of AI through the way that all of the data and models in the world can’t actually convince people with power to make decisions that would actually save all of our lives, but to in fact do the opposite.

19431

Reposted by Benjamin Borowski 🍁

neutral @neutral.zone · 22/09/2026

Bounded authority, observable decisions, revocation, rollback, containment, separation of duties, liability, human escalation, explicit handling of unknowns. Basically all the boring machinery civilization eventually invented precisely because we gave up on people reliably being aligned.

04211

Benjamin Borowski 🍁 @typeoneerror.com · 22/09/2026

Unless you are quite young, that seems pretty minor to me. But of course, I'm not a medical professional (beyond my first responder license). I take guanfacine which helps with bp and adhd. They'll tell you to exercise, but stress/poor sleep has always been the big one for me.

110

Benjamin Borowski 🍁 @typeoneerror.com · 22/09/2026

I'm confused. I have high blood pressure and I would love to get 116?

110

Benjamin Borowski 🍁 @typeoneerror.com · 22/09/2026

babysitting a friendo for the week. we need a dog again.

051

Benjamin Borowski 🍁 @typeoneerror.com · 22/09/2026

Other topics: - Generating Notion workspaces w/TypeScript - Building mini apps using bound data with Custom Blocks - Connecting custom tools with Custom Agents - Securing workspaces for multi-player Notion AI use (though I'm not shy about saying I don't use Notion AI: my goal is harm reduction).

Notion AI can be quite powerful for organizations and individuals. It can give similar “skills” to everyone that traditionally only the programmer/nerd-types like myself used to have. Which also means that now everyone can have very powerful editing and creating tools at their fingertips. Lots of folks are finding their available “skill-sets” increasing daily as they discover more about how these tools work.
That being said, “Agentic AI” also makes it quite easy to:
Accidentally make changes to your data you didn’t intend to.
Get flawed information from the systems that impact decision- and sense-making.
Create extremely noisy data streams that create an abundance of busy work.
Quietly expose your confidential data to third parties.
If you decide to leverage Notion AI, we want to minimize these outcomes, and a lot of what we’ve

020

Benjamin Borowski 🍁 @typeoneerror.com · 22/09/2026

Building a series of Workers as examples. We've been showing students how to connect custom built tools to agents, and using oauth, sync, and webhooks to manage things from and outside of Notion. Example: my system for managing my SoundCloud -> syncing track metadata, metrics, managing playlists.

A notion interface showing a bar graph with a series of "plays" for tracks

A gallery view in Notion showing a series of album covers with titles, icons, and a link to soundcloud

100

Benjamin Borowski 🍁 @typeoneerror.com · 22/09/2026

Working on October's "Coding with Notion" workshop. Many of the community are on board with coding agents, but I've seen (over numerous private workshops) a lot of folks are—even with these "skills"—struggling to finding their way though the developer platform, so I'm doing a "non-AI dev" series.

220

Benjamin Borowski 🍁 @typeoneerror.com · 21/09/2026

Still not enough transparency in the way modern "AI" products are developed, but perhaps this kind of MicroHype is a path toward transferring HypePrime away from chatbots.

000

Benjamin Borowski 🍁 @typeoneerror.com · 21/09/2026

word, I think you're spot on. I think @colin-fraser.net said as much just. thanks!

000

Benjamin Borowski 🍁 @typeoneerror.com · 21/09/2026

Appreciate your answer! I'm considering that a pre-trained, non-autoregressive model might be similarly "seductive," especially knowing how the average LLM-user will use a tool like this. But the target of the seduction is at the software decision-making layer: as concerning in some aspects.

110

Benjamin Borowski 🍁 @typeoneerror.com · 21/09/2026

After playing around with it and getting a feel for its use-cases, I'm curious what you mean by "jev can't seduce anyone." Say more?

100

Benjamin Borowski 🍁 @typeoneerror.com · 21/09/2026

ISP was down most of the weekend so I ended up finishing the flooring and baseboards in the closet, which I was planning on putting off for ~5–10 years. Better than calling Telus this morning, anyway. Let's update my "Guide to the 10th, 11th, and 12th circles of Hell, or Telus Corporation" guide.

# Account details

The product we are paying for is called “Business Wireline.” You will not find this product name on the website, your account, nor your invoices.

**Business Wireline:** 1-800-361-3311

When you call Telus you will be fed into their AI assistant. Ask to “talk to a human” to be forwarded to the *general* support queue. Immediately declare “please direct me to the support line for Business Wireline”.

<aside>
⚠️

Note that the phone number below will feed you into this AI loop and there apparently is no longer a direct support number for business customers.

</aside>

**Service number:** 604-740-0137

<aside>
⚠️

*Note: our account has a phone number associated with it that is not our actual phone number! I have not had success with getting them to change it. Isn’t this fun?*

</aside>

000

Benjamin Borowski 🍁 @typeoneerror.com · 21/09/2026

Man, I'd love if my entire feed were people just opining about rescue. We don't discriminate: underwater, fast water, rope rescue, confined space... Please. Talk to me.

010

Benjamin Borowski 🍁 @typeoneerror.com · 21/09/2026

I don't think there's a perfect way to fully do away with incarceration for the most extreme of humans, but I still feel really great about arguing for it, because not arguing for it seems to me to be a great way to maintain the quo of for-profit private institutions.

110

Benjamin Borowski 🍁 @typeoneerror.com · 19/09/2026

I think often there's a value in giving dangerous things "agency" as a reminder to handle them with care.

000

Reposted by Benjamin Borowski 🍁

vortex_egg @vortexegg.com · 19/09/2026

An interesting thing about AI, that we've classified under everything from hallucinations, to slop, psychosis, and unaccountability, is that it is creating a virtual world of appearances without content. The superficial seeming of things, and people acting on them as such, with no there there.

69616

Benjamin Borowski 🍁 @typeoneerror.com · 19/09/2026

masterminds are an underrated peer group format. so is volunteering, but I'm oppressive with trying to get people to become volly firefighters. heh.

010

Benjamin Borowski 🍁 @typeoneerror.com · 19/09/2026

it's going to take me awhile to read RSI as AI-related and not the reason I had to learn how to mouse left-handed years ago.

000

Benjamin Borowski 🍁 @typeoneerror.com · 19/09/2026

My friday wfh: Coffee + writing. Then a fire call: a very funny 92 y/o patient. made some music for an hour, followed by a consulting call, a live teaching session, then another 911 call: this one a "wait for police". Now I'm having a nap and playing some Balatro. Very inconvenient, very vibrant.

031

Benjamin Borowski 🍁 @typeoneerror.com · 18/09/2026

Guess this is why people keep saying "the bottleneck is discernment and judgement." When you can "just do things," you will. Everyone's terrified their "thing" is token-commissionable, so perhaps we're struggling with commitment.

020

Benjamin Borowski 🍁 @typeoneerror.com · 17/09/2026

man, that hyrox "a win is a win" lady is really the perfect metaphor for the modern individualist progress narrative: i don't care who has to eat my shit as long as i get mine. the fact the company initially canonized it as perseverance is abs demented.

110

Benjamin Borowski 🍁 @typeoneerror.com · 17/09/2026

I do like the idea of reactive models that don't generate at all though, where opportunities are decided and a model guides it towards "typed" outcomes. feels much more like the "discernment/judgement" that LLM fans say is paramount. With "rights clean" models, this tooling is quite interesting.

000