HN user

michaelje

160 karma
Posts5
Comments48
View on HN

Once upon a time, physically shipping faulty software had real costs borne by the organization - production, redistribution and transportation of a physical disc.

Today there’s no disc, no recall - that cost to shipping broken software is gone. We the users pay the price.

Open models keep closing the eval gap for many tasks, and local inference continues to be increasingly viable. What's missing isn't technical capability, but productized convenience that makes the API path feel like the only realistic option.

Frontier labs are incentivized to keep it that way, and they're investing billions to make AI = API the default. But that's a business model, not a technical inevitability.

Absolutely. Pricing exposure is the quiet story under all the waves of AI hype. Build for convenience → subsidise for dependence → meter for margin is a well-worn playbook, and AI-dependent companies are about to find out what phase three feels like.

Hyperscalers are spending a fortune so we think AI = API, but renting intelligence is a business model, not a technical inevitability.

Shameless link to my post on this: https://mjeggleton.com/blog/AIs-mainframe-moment

https://sorso.app/

A fresh PWA to log / improve your coffee brewing process. We use it to see what we are all drinking, find new coffees, explore new cafes, and understand what we like / don’t like.

It’s primarily used by our group of friends, so if you see a rough edge somewhere please reach out!

> critics argue that it reduces the supply of rental apartments and creates housing shortages

I don’t follow the logic. It doesn’t destroy the property or take it out of supply, it just means someone might not purchase it specifically to rent it out. Which in turn means less demand buying houses overall, which means more opportunities for people to buy (if they want?)

Agree - I’ve been using a set of Eve (from memory) power plugs that are thread enabled - my understanding is it doesn’t connect directly to the internet (ie you don’t connect it to your wifi) but you connect to a border hub (ie an Apple TV, or maybe now an iPhone 15.)

So I guess if you trust apple, you can sorta trust these devices as their access to the outside world would be through an apple device.

I just got back from the Banff Film Festival, where I saw Fire of Love[1]. It’s the story of French scientists Maurice and Katia Krafft, who become fully devoted to each other and the study of Volcanos for the rest of their lives.

Created using hours of their archival footage, it’s a story driven by a singular shared passion, amazingly found by two people born only 20km apart.

[1] https://youtube.com/watch?v=kz95Xjd3l00

I’ve just completed a 6 month self comparison of the Whoop v4 and Watch S6 for cycling. I found for long endurance efforts they were surprisingly accurate to each other - but during intervals of intense exertion (eg a KOM/hill climb) often the Whoop would read my HR as ~40-60BPM lower while the watch would reflect an expected reading (160-180bpm). Sometimes after 1-2 mins the whoop would “catch up” but it would leave a giant drop in HR graphing for that interval.

This also lowered the avg HR for the workout on the whoop, as you noted. Happy to share an example comparison graph if you’re interested, just reach out.

For what it’s worth, I ended up cancelling the Whoop this month after trying twice to engage with their data team.

How can we do better to see, stop, or price-in externalities that impact people, communities, the world?

This is a widespread issue across many essential areas of human life - like housing (Airbnb), or the environment (pollution).

Please don’t get less than 16.

I have now had two (work and personal) - my work m1 air was specced with 8GB, and had major memory pressure and issues with an external monitor. My personal 14” MBP has 16GB - same workload, same monitor, never skipped a beat.

Ended up sending the air back and swapping for one of the old dev ex MBP (intel based).

No - it has to be more casual.

I don’t want to force people to join a/my/some club. I want to let my followers and friends see I plan to do a 80km ride Saturday morning and let them simply tag along. It could easily be integrated into the feed or as a separate section.

I’d argue social activity is one of the biggest opportunities for Strava to lock themselves in as the hub of all movement activity - but it requires supporting how people like to naturally organize in a local community.

What are the key differentiators to get me to use this over Strava where people already are? A quick look and some MTB specific metrics like hangtime?

A killer feature for Strava (or anyone in the social activity game) would be allowing me to setup social rides with my followers/friends. For example, I plan on doing this loop at 8am Saturday - and anyone who follows can hit ‘+1’ and let me know I’ll join. Where I live, everyone does similar road rides, but won’t coordinate times which would be a great social opportunity.

There’s no messaging or ability to setup casual rides - and if community is a key pillar for them, it’s a huge way to move further up the activity stack to planning activities (which they already have some basic functionality for mapping and planning routes for yourself).

As another poster mentioned, there are privacy considerations but make it opt in and if implemented well would be a killer feature to lock in community as a pillar (and moat). I’ve been thinking about this a lot lately. FB is trying to do this with community on WhatsApp. Someone please eat their lunch. Anyone.