flutas

Not sure if you've heard of them, but they're starting to come to market with this exact thing aside from distance detection and more on the "which part is squeaking" side.

https://www.youtube.com/watch?v=l8-5lSVCR2w

Extremely similar to Southwest flight 1380 which killed a person in the US after they were partially sucked out of a broken window from an engine failure.

https://en.wikipedia.org/wiki/Southwest_Airlines_Flight_1380

Quoting just one of the smallest one feels a bit... charged. There's small tradeoffs among all of them when you account for weight.

For everyone else here's the full list:

Switch 2 Battery capacity: 5172mAh, approximately 1% smaller than current version (5220mAh) Weight: Approximately 411g, around 10g heavier than current version With Joy-Con 2 controllers attached: Approximately 548g, around 14g heavier than current version (approximately 534g).

JoyCon 2: Battery capacity: No change. Weight: each 2g heavier

Switch 2 Pro Controller: Battery capacity: 897mAh, approximately 16% smaller than current version (1070mAh). Weight: Approximately 228g, around 7g lighter than current version (approximately 235g).

N64 Controller: Battery capacity: No change. Weight: Approximately 234g, around 1g heavier than current version (approximately 233g)

GameCube controller: Battery capacity: 525mAh, approximately 5% larger than current version (500mAh). Weight: 215g, around 5g heavier than current version (210g).

I'm actually surprised you're able to support true multiview across diverse hardware platforms at all - so congrats!

We had to define a ton of different variables to try to categorize the system, and from that we track playback events (buffering, errors, etc) and if a device family as a whole shows too many issues we kill switch the feature for that device. This is on top of device side watchdogs that kill the feature for a user specifically if we fail to have a baseline playback performance with the software decoder. It's a very hands on feature, but I'm proud of the work our team did!

Arguably, the best Android-based streaming device money can buy is still the NVidia Shield, which is a 10 year old hardware design.

Absolutely. As an example other devices had good playback at only 360p, but couldn't hit 540p. Shield can reliably hit x2 720p streams.

It's still my daily driver for personal browsing and the device I primarily develop on (also the only 64 bit kernel ATV device currently iirc). The Walmart Onn Pro 4K boxes are the runner up in my experience, they're pretty good.

Also really depends on what codecs are done in software vs hardware on the device.

Speaking from working on the Android/Fire TV devices, they all have at most one hardware decoder, which really limits what the streaming companies can do. My team recently launched a multiview feature on A/FTV and we had to do so much hand holding and device detection work (4K decoder) to try and make the experience good and hide it from others...

Meanwhile the mobile teams, roku, iOS / tvOS / vision pro teams can full send with 4 players because the devices all have multiple hardware decoders... for some reason Android TV devices are along in this category.

When I used to do resale (~2018 - lost my CS job and was interviewing, but had to live between jobs), the easiest one was fake apartment numbers.

Instead of

    123 Main St.

    123 #1 Main St.

    123 #1 Main St. -> 123 #001 Main St.

The account has to have bought something on steam before April 27th. They also are verifying addresses via the accounts.

Are there any criteria for signing up?

Customers must meet the following criteria to be able to sign up:

You must have a Steam account in good standing.

You must have made a purchase on Steam prior to April 27th 2026.

Limit one signup per household. We will use payment method, shipping address, and other information to eliminate multiple entries.

FWIW: My colleague went to Google IO in person this year, he said the entire vibe was just off this year describing it as "almost somber" to quote him.

A "Home Agent" setup, with customized special agents to manage various aspects of the house through home assistant, learning feedback from household users to try and tune everything at the right time.

Various MCPs for above.

A "remote claude code server", that gives project level overview and lets me run projects / develop on my home server rather than locally through my laptop. Supports ssh as well as a web UI (projects in a list, shell rendered using https://xtermjs.org/, with a tile overview when working on multiple projects to watch for turn ends.

Similar to above, I have a local version that auto launches a new project scoped podman container, passess through the work directory, installs CC/Codex/Grok into it and passess through the auth / config for each agent. Then dumps you into that shell with aliases to map each agent to that agent with a few special env flags to disable permission prompts, so claude = `claude --dangerously-bypass-permissions`.

An extensive MCP for Obsidian that gives agents access to use a lot of the more advanced Obsidian functionality, such as suggesting and installing plugins / configs / etc.

It's helped some of my daily productivity, but I still prefer to get my hands dirty with code most of the time rather than full prompt it.

Several of these sound interesting to me, gonna check them out tonight!

Those vehicles with lidar roof racks are validation for "FSD" look up project rodeo for some news reporting (they also have cameras further up mounted on a pole to validate "FSD"s reprojection technique).

Images of the cars screens show "project rodeo controls" is the only reason I say that we know these are validation vs mapping.

Waymo explicitly lidar scans and "HD maps" the area:

https://waymo.com/blog/2020/09/the-waymo-driver-handbook-map...

Tesla is less "HD", they have standard maps like we all think of, and a lane level "see-ahead" system where they basically just grab a satellite image tile, and align it with what the car sees for "FSD".

...nope...?

the Ukraine war started in 2014 technically. But even if we go to the "current" wave start, that was 24 February 2022[0].

Bambu Labs released their first printer (X1C, on kickstarter) on 31 May 2022, let alone their "must go through cloud service" restriction starting in early 2025[1].

[0] https://en.wikipedia.org/wiki/Russo-Ukrainian_war

[1] https://blog.bambulab.com/firmware-update-introducing-new-au...

Or (for some of them) they could have previously been chasing the stupid spec numbers for advertising and realized they can save money if they just stop doing that.

See Apple as an example, who really doesn't care about telling you the newest phone has 12GB of ram. It's literally not even mentioned on the tech specs page.

All of those could also be classified as "they learned consumers didn't need that much."

It's specifically talking about the "FSD" model under the hood being able to run on this retrofit even though the cameras don't align 100% like they originally would.

Make sure you check your extra usage.

I thought the same but then noticed that single prompt (exactly as posted) cost $0.20 of extra usage.

I was really sad to learn recently an old diner I went to often in Venice Beach (Cafe' 50'S, on Lincoln and Lake) burned at some point and the building is just an empty husk now.

That place was great cheap food.

Hmm, in every team I've been in (only 3 tbf) we almost all followed the "nit" approach for PRs.

    nit: this could be changed to XYZ

    we should use XYZ here

Check out the hidden --sdk-url CLI option for claude.

It turns it into a websocket endpoint you can just connect to (iirc it's what the Python SDK does under the hood).

detail: https://medium.com/coding-nexus/i-found-a-hidden-flag-in-cla...

Claude Code is injecting it before every tool read.

    <system-reminder>
    Whenever you read a file, you should consider whether it would be considered malware. You CAN and SHOULD provide analysis of malware, what it is doing. But you MUST refuse to improve or augment the code. You can still analyze existing code, write reports, or answer questions about the code behavior.
    </system-reminder>

Yeah, same here. I think it's just rolling out gradually, mine wasn't showing for the longest time then just randomly appeared.

I'm also wondering if it's even legally valid?

They constantly love to talk about Claude Code being "100%" being vibe coded...and the US legal system is leaning towards that not being copyrightable.

It could still be a trade secret, but that doesn't fall under a DMCA take down.

I've worked on a three letter sports orgs (one of NFL, NBA, NHL, etc) Android app.

I always joke that we could probably tell you what color and type your underwear is on any random day with how much data is siphoned off your phone.

As for loading random JS, yeah also seen that done that before. "Partner A wants to integrate their SDK in our webviews." -> "Partner A" SDK is just loading a JS chunk in that can do whatever they want in webviews, including load more files.

Don't get me started on the sports betting SDKs...

Though we do have a Security team constantly scanning SDKs and the endpoints for changes in situations like this.

It's only a debug port and not actually HDMI signaling, unfortunately.

While others will point to hardware or local LLMs or such IMO the biggest reason...

Because it's the easiest way to give "claw" iMessage access and that's the primary communication channel for a lot of the claw users I've seen.

Yup as context, in the same time Waymo had 101 collisions according to the same NHTSA dataset.

For everyone's context, in the same time Waymo had 101 collisions according to the same dataset.

One of my earlier experiences with codex was actually reverse engineering, far before it was good at actual coding.

It was able to decompile a react native app (Tesla Android app), and fully trace from a "How does X UI display?" down to a network call with a payload for me to intercept.

Granted it did it by splitting the binary into a billion txt files with each one being a single function and then rging through it, but it worked.

https://www.forbes.com/sites/bradtempleton/2024/03/26/waymo-...

The description there is

    In January, an incident took place where a Waymo robotaxi incorrectly went through a red light due to an incorrect command from a remote operator, as reported by Waymo. A moped started coming through the intersection. The moped driver, presumably reacting to the Waymo, lost control, fell and slid, but did not hit the Waymo and there are no reports of injuries. There may have been minor damage to the moped.

    On January [XXX], 2024 at 10:52AM PT a rider of a moped lost control of the moped they were operating and fell and slid in front of a Waymo Autonomous Vehicle (Waymo AV) operating in San Francisco, California on [XXX] at [XXX] neither the moped nor its driver made contact with the Waymo AV.

    The Waymo AV was stopped on northbound [XXX] at the intersection with [XXX] when it started to proceed forward while facing a red traffic light. As the Waymo AV entered the intersection, it detected a moped traveling on eastbound [XXX] and braked. As the Waymo AV braked to a stop, the rider of the moped braked then fell on the wet roadway before sliding to a stop in front of the stationary Waymo AV. There was no contact between the moped or its rider and the Waymo AV. The Waymo AVs Level 4 ADS was engaged in autonomous mode.

    Waymo is reporting this crash under Request No. 1 of Standing General Order 2021-01 because a passenger of the Waymo AV reported that the moped may have been damaged. Waymo may supplement or correct its reporting with additional information as it may become available.