Same content, different presentation. Open the Capacity Controls (bottom-right) and drag the sliders — or pick a preset like Exhausted or Energized — to see the feed restructure in real time.

Showing a curated selection — increase cognitive capacity to see all posts

@elaravoss4h

There's a specific kind of grief that has no name, because naming it would be embarrassing. It's the feeling you get when someone you follow online does something disappointing. It's not quite betrayal—you never had a relationship. It's not quite sadness—you didn't lose anything concrete. It's a hollow confused ache that has no proper home.

We've been rewiring our emotional expectations to include people who don't know we exist. Creators, newsletter writers, podcast hosts—anyone who broadcasts consistently into your brain three times a week for years. The intimacy that accumulates is real, in the sense that real feelings accumulate. The relationship, however, is entirely unilateral.

What worries me isn't that parasocial dynamics exist. They always have. What worries me is the texture. Before, parasocial feeling was reserved for rockstars and movie stars—a category of person you understood to be remote by design. Now it's someone who records themselves in their bedroom and calls their audience "the community" and responds to comments and knows your first name.

The intimacy architecture is indistinguishable from real friendship. The brain can't tell the difference. And when the disappointment comes—and it always comes, because people are complicated—we have no script for processing it. The grief is entirely unprocessed because we can't even name what we're grieving.

@mchen_ux7h

Every major product team is solving the same problem: how do we make users return? The answer every team has landed on is identical: become a destination. Add features. Surface more content. Give people more reasons to stay.

The result is that every tool wants to be a platform. The note-taking app wants to be your second brain and collaboration hub and publication tool. The project manager wants to surface your calendar and email and team wiki. The photo editor wants a community feed and a marketplace and learning videos.

I don't think this is cynical. I think most teams genuinely believe more is better. But I've started to notice that the most hostile interface design is refusing to let you finish something and leave.

The apps that actually improve my life are the ones that don't need me to return. They do their job and let me go. They're increasingly rare, because there's no business model for "helps you accomplish something in four minutes and then you leave satisfied." The business model is time-in-app.

The tool that respects your time is the tool that makes itself easy to ignore. Nobody builds for that.

@tkl_writes11h

Libraries are having a crisis of legitimacy precisely when they're most needed. We're drowning in information—most of it low quality—and the institution specifically designed to help people navigate it is being chronically defunded while the information keeps multiplying.

The argument for cutting library budgets is almost always "information is free online now." This misunderstands what libraries actually are. Libraries are not about access to information. The internet handles access. Libraries are about trusted curation, professional information literacy, and a physical space for people who need quiet and help.

When someone can't distinguish a PubMed abstract from a wellness influencer's post about the same topic, that's an information literacy problem. When a student can't evaluate whether a source is reliable, that's a curation problem. When someone is job hunting and needs help with their resume but has no printer and no private space, that's an access problem the internet didn't solve.

Libraries solve all three. We're defunding them because we've confused "the internet exists" with "information problems are solved." The internet created more information poverty than it solved. You can have access to everything and be completely unable to find what you need. We built the indexing layer without the interpretation layer, and libraries are the interpretation layer, and we're cutting them.

@ossdaily2h

The EU AI Act's first major enforcement cycle began this week, and the gap between what open-source advocates hoped for and what the legislation actually requires is becoming clear.

General-purpose AI model providers above the compute threshold (10²⁵ FLOPs) must comply regardless of release format—open weights included. The exemption that advocates argued for, covering models released under open licenses, only applies to providers below the threshold. All major frontier model labs and several open-weight labs are above it.

The practical implications: providers must maintain technical documentation, transparency reports, and conduct adversarial testing before significant capability updates. Models trained on EU data have additional requirements around data governance. The first compliance deadlines hit August 2026.

Early signals suggest the major US-based labs are moving toward compliance; several open-source focused organizations are still working through what compliance looks like without the revenue streams that make it tractable.

@climate_brief1h

Weekly climate metrics:

Atlantic sea surface temperatures remain at 3-sigma above baseline for the fourth consecutive month. The anomaly has been persistent since February and is the longest such streak in the satellite record (1981–present).

Arctic sea ice extent is tracking below the 2012 record pace for the first time since 2020. Current extent: 4.2M km², against a 1981–2010 average of 5.9M km² for this date.

Global mean surface temperature anomaly for Jan–June is +1.67°C above pre-industrial baseline (Berkeley Earth). This would be the highest January–June anomaly in the instrumental record if it holds.

Extreme heat events this week affected approximately 800M people across South Asia and the Middle East, with wet-bulb temperatures exceeding human thermal tolerance in parts of the Gulf for the third consecutive week.

@juniperro30m

the tomatoes are finally blushing. been talking to them every morning for six weeks and it worked, or just the sun did. either way.

@devdispatch3h

hot take: the README is the product. everything else is implementation detail.

@nadiaokonkwo6h

thread: types are documentation that the compiler checks. everything else about them is secondary. going to try to convince you of this in 12 posts.

23 replies in thread

6 more posts — increase cognitive capacity to see all