Ricardo J. Méndez
banner
ricardo.bsky.social
Ricardo J. Méndez
@ricardo.bsky.social
Privacy, decentralization, freedom, the odd crypto post.

Career CTO working on strategy and investments.

https://numergent.com/
Pinned
#Monkigras is not a conference - it's a communal hallucination where for two days we believe that talks can be creative, conversations can be individual and hustle-free, event food can be great, and corporations can help with inclusion.

You should come!

monkigras.com
Reposted by Ricardo J. Méndez
I think it's wrong to describe the World Service as "soft power". It's actually "info power". And these days that's a very hard form of power indeed.
February 15, 2026 at 8:05 PM
Ok I really need to do a write up on the latest version of my network reputation system, and how it would catch closed loops like this one.

Maybe on Wednesday.
I’m pretty confident in calling this fake. The user who posted it seems to be part of a substantial network of accounts with similar usernames all posting and commenting on each other’s content. Ironically it’s a great example of an AI agent making stuff up to tell you what you want to hear
February 15, 2026 at 12:45 PM
I pride myself on trying to process things from someone else’s view but this is beyond me. How do you decide this is a hill for you to even approach?

Ignorance from your audience on who Epstein was? Raw tribal signaling “thought leadership” value of taking the most absurd position possible?

🤯
Top to bottom, an incredible image
February 15, 2026 at 8:36 AM
Reposted by Ricardo J. Méndez
the two types of vehicle are cars and trucks, with every vehicle being a subclass of one of those two. a jeep is a truck. a motorcycle is obviously a car. a boat is a truck. an airplane is a car and a helicopter is a truck. trucks are cars, interestingly enough, and cars are a specific type of chair
February 13, 2026 at 6:06 PM
Reposted by Ricardo J. Méndez
Yes there’s a wine app, @winesky.app! Follow wine expert & app builder @dr.h.winesky.app for more.
Marqués de Cáceres Verdejo by Bodegas Marqués de Cáceres (2024) - 4 ⭐

We chilled the wine quickly and even added few ice cubes. Not too dry not too sweet, pleasant acid especially with some cured meats and aged cheese.

White Wine from Castilla y León

via @winesky.app

🍷 #winesky #spanishwine
February 9, 2026 at 10:56 PM
“I can’t tell you what we are doing, but it’s cool and above board - if you want to know more, just sign this thing saying you won’t ever tell anyone how above board it is.”
SCOOP: Tensions at Palantir have grown in recent weeks over the company's work with ICE.

On Friday, CEO Alex Karp tried to calm concerns in an hourlong video that offered few specifics—instead offering NDAs to workers who want to understand how it's empowering ICE.
www.wired.com/story/palant...
Palantir CEO Alex Karp Recorded a Video About ICE for His Employees
In a video shared with Palantir employees, Alex Karp did not explain how ICE is utilizing the company's products. Instead, workers were told they can sign NDAs if they want detailed information.
www.wired.com
February 10, 2026 at 8:33 PM
What he is saying seems to be: don’t cut off your nose to follow your discourse.
it is already possible, right now, to get 100tok/sec out of 100 watts of air-cooled local inference, and the output is surprisingly good

LLM architecture gains will compound against improved silicon over the next five years to produce serious leverage, WHICH WORKERS CAN OWN OUTRIGHT
If you're a person who thinks "billionaires are evil -> billionaires own big tech -> frontier labs dominate the LLM space and are owned by billionaires -> LLMs is evil" then I am sympathetic to it. However the correct prescription is not to destroy LLM technology but to expropriate it.
February 8, 2026 at 10:08 PM
Am I imagining this, or did @github.com removed the option where individuals could enable or disable specific models?

I have a subscription on my personal account, and can now only find the setting for the org I created for a couple of private repositories.

"Auto" on Copilot Chat is now useless.
February 6, 2026 at 4:26 PM
I mean, if only someone had warned us 3 years ago that Ethereum Layer 2s were in practice uninvestable...

numergent.com/2023-08/No-L...

Or that it was unlikely to re-capture developer imagination because of fucked-up incentives...

youtu.be/6VWOKEMAD9A?...

Or... (I could keep going...)
February 4, 2026 at 8:14 PM
Does tap's database just keep growing, or does it at some point start culling old, processed records to merely keep some cursor or a "processed" indicator instead of all the data?

#atdev
February 3, 2026 at 7:24 AM
Financially… for Google, $68M is such a slightly sighed breath atop the financial bucket, that it might as well be read as “we ain’t doing it but we’d rather stop paying lawyers”.
Do you remember when people were saying "I think my phone is listening to me, i'm getting ads for stuff I spoke about" and people said "They arent listening to you, they're just this good at predicting your interests"?

Well, it happens they were doing exactly that.
www.cbsnews.com/news/google-...
Google to pay $68 million over allegations its voice assistant eavesdropped on users
Class-action lawsuit alleged that Google's voice assistant illegally recorded and shared private conversations with advertisers.
www.cbsnews.com
February 1, 2026 at 11:12 PM
Absolutely bananas. I suspect we have entered the era where we'll see a lot more vertical take-offs like moltbot's, along with rugpull-like plunges as they get fast-replaced.

Things are going to be lively for a while.
bonkers from Peter Steinberger. moltbot hype is literally off the charts.
January 30, 2026 at 8:28 AM
Global Latin America strikes again.
I don't really like how habituated I've gotten to blatant corruption and abusive behavior
January 29, 2026 at 3:37 PM
Five jobs I’ve had:
- Brick factory gopher
- Software engineer (all along the spectrum)
- Movie critic
- Investor
- Used bookstore clerk / janitor
Five jobs I’ve had
- Grocery store
- RadioShack
- Electrician
- Community Organizer
- Software Engineer

Bonus:
- atprotocol button smasher
Five jobs I’ve had
- Grocery store/meat department
- Construction
- Little Caesars
- Desktop Support
- Software Developer

Bonus:
- atproto pumpkin
January 29, 2026 at 3:01 PM
Deleting Signal now that I know it’s compromised. Find me atop a wooden soapbox in the market square, screaming in tongues.
Deleting Signal now that I know it's compromised. Find me yeeting every message I wanna send into a single @leaflet.pub doc (one subpage per recipient).
Deleting Signal now that I know it's compromised. Find me on guestbook.pl
January 28, 2026 at 10:59 PM
Reposted by Ricardo J. Méndez
January 28, 2026 at 4:36 PM
Chronicle of a Death Foretold for anyone who has dealt with NI or their products in the last 7 years - and that's before we get to the strip mining-smelling private equity acquisition , general state world finances, and how every geek and their great-aunt now build synths.
January 28, 2026 at 2:52 PM
Soon I'm going to have to split my "ATProto Dev" list into specific areas.
Excited to announce that Aeronaut is now live in the Mac App Store! 🎉

After two years of tinkering, polishing, and obsessing over the little things, it’s finally here!

If you’re looking for a Bluesky app designed and built for macOS, take a look…

apps.apple.com/us/app/aeronau…
January 28, 2026 at 9:31 AM
We can just do things, IRL.
Just a dude doing things
January 27, 2026 at 2:28 PM
This is where 2B and 9S hangs int for coffee.
the cafe in the basement of the hidden library
January 26, 2026 at 7:19 PM
3000 posts later, @xkcd.com 386 remains undefeated.

xkcd.com/386/
Duty Calls
xkcd.com
January 25, 2026 at 3:27 PM
Solid read here from Cat. She doesn't go into my main reason not to build on it, though (besides of revulsion at Altman's sleazy, bullshit, sociopathic snake-oil salesmanship, and against the advise of friends who are better at growth than myself).

I've seen this movie before.

1/
OpenAI isn't just building helpful tools. They're building a ladder.

ChatGPT Health launched three weeks ago to significant fanfare. 230 million people already ask ChatGPT for medical advice, so a dedicated health product makes sense, right?

Well, I have some thoughts in my latest post.
The Ladder to Nowhere: How OpenAI Plans to Learn Everything About You
ChatGPT Health is a small part of a much larger plan to learn everything about you. In this post, I talk about what's driving them and how they might get there.
insights.priva.cat
January 25, 2026 at 12:46 PM
Whoodathunk! BSC does have users after all!
My latest blog post investigating a malware campaign which infects victims by utilizing only legitimate infrastructure. The malicious activity spans hundreds of hacked websites, the BSC blockchain, and a popular CDN.

expel.com/blog/clearfa...
ClearFake gets more evasive with new living off the land (LOTL) techniques
ClearFake's latest campaign uses fake CAPTCHAs and social engineering trick victims into installing malware, and it's getting more evasive.
expel.com
January 25, 2026 at 9:33 AM
Gave specstory a quick try - I might be missing something, but it doesn't seem that useful for me.

If what I want to track is the why of a change, then good ol' issue notes, along with using spec-kit and manually modifying the first pass docs it generates seems a much better approach.

1/
January 24, 2026 at 9:07 AM