jonpurdy

That's good to know. I would assume if Apple actually released it they would have ensured the hinge would be as reliable as possible (of course, Hingegate is always a possibility).

I'd also go for an iPad/Surface form factor with detachable keyboard if I had to.

The only reason I would want touch is if the screen would fold and rotate in order to turn it into a tablet. If Apple made this in a 12-inch format (like their old 12-inch MacBook I'd buy it immediately.

No idea why anyone would want a laptop form factor with a touch screen; seems like an RSI nightmare. Especially since you can already pinch-and-zoom and do other gestures with the existing touch pads!

I bought my 2022 M2 Air with 24GB/1TB/10-core GPU. Very very glad I did.

The only reason I’m bitter is because they put in WiFi 6E in the M3 generation. Now I just use a 2.5Gbe USB-C adaptor when transferring large files to and from my NAS.

In Korean, dogs say meong (멍), sounds like mung) rather than woof.

I don't mind at all. So I had ~100 subscriptions as of a few years ago and I tend to just look at the Subscriptions page, view what's been released since the last time I checked, and download those.

I sometimes see the homepage, or see thumbnails when I click to download individual videos. I've previously found interesting channels that way, or recommended by creators in videos I watch, or even on HN.

(I've found their algorithm has gotten worse over time, and nosedived recently with AI slop propagating like crazy in 2026. Even for niche topics it'll recommend channels that are a month old, already have 60 videos (mostly with low views but one video with 100K+ views), and obviously AI generated when closely inspected. It's worse now because half the channels use AI thumbnails for non-AI content.)

So after that long answer, I don't have a great recommendation for discovery if you don't already have an existing list of channels you like and trust.

Yes. YouTube still thinks my partner and my accounts are in different countries so we couldn’t get Premium Family (for the past 4 years) and I’m not about to purchase separate subscriptions. So I have a backend that I can send YouTube URLs to running on my Mac Mini that gets the videos with yt-dlp.

I've literally been waiting since 2022 for this.

I purchased the 2021 ATV just before the 2022 was released. It struggles with AV1 and VP9 (Infuse really struggles, VLC is better but still a problem). I just wanted the faster CPU (for software decoding AV1 and VP9). Almost bought the 2022 version in 2024 for that but didn't want to buy a 2 year old product.

Didn't expect to have to wait 4 years, and now the price is more than doubled. I can afford it but refuse to do so, especially since I don't use any of the onboard storage for anything.

I personally don't mind the multiple interview slots or 9 hours of conversations. It's fair since it takes their team members an equivalent amount of time (though it's split across team members, each team member is likely interviewing many others as well).

My blocker has been the initial written answers to the questions. I've had the questions half-answered for nearly a year now. I want to do it properly (sans AI) but other things come up and I question the EV of spending the time, especially since I'm not a strong writer in the first place※.

For Oxide, it makes sense to demand a strong writer due to their culture, so even if technically qualified I might not be a good fit, assuming I even get an interview.

※ - I write unambiguously, but use point form whenever possible and am not great at naturally writing concisely.

As a reference point, I’ve stubbornly been using A4 paper in both USA and Canada for years now with governments and businesses. Never had any problem with IRS, passport agency, CRA refusing to accept them, or even mentioning it at all in person.

No problem buying it either, though it’s about 2x the price of letter. (I go through about 100-200 sheets per year mostly due to my daughter drawing on it.)

I don’t actually store any printed paper since anything gets scanned and shredded, so no comment on long term filing.

(Due to this experience I recently partially switched to B5 paper, which is a nice size compromise instead of A5 which would be too small. Anything A4 will print to it without scaling issues of course, just takes up a bit less space.)

Controversial take but this works better for me than just the weekly limit. I have about 1-2 hours of sustained focus per 5 hour period so if I run out of tokens then I can take a break and start again in a few hours. Or just use the top ups that I’ve paid for in case I want to push through.

Having just the weekly limit meant I could blow through too much within the first couple of days. Fortunately, resets were raining down during that time.

(I know I could just vibe code a tracker that splits up my usage into arbitrary periods and lets me know when to take a break. Maybe if they remove the 5-hour limit again.)

Also, for files in a directory with the same prepended string with long filenames(like full_show_name - S01E01), it displays just a fraction of the full filename in a list.

I need to sit there and wait for the filename to scroll horizontally to see what the actual episode number or name is. Instead of just showing the full filename on multiple lines, or showing the filename below the thumbnail.

Also:

- Flattens folders with a single file in it (so foldername/file.mkv shows up as just file.mkv in the parent folder), no way to disable this (https://community.firecore.com/t/avoid-flattening-tv-show-fo...)

- Doesn't show landscape artwork (embedded .png), only portrait, and if the embedded artwork is not portrait it grabs a random frame from the file instead of cropping the embedded artwork (https://community.firecore.com/t/use-landscape-artwork/26878...)

- If you exit a video before it's ended, but it's close to the ending, it marks it as played (so if you go back in you lose the playback position) (https://community.firecore.com/t/infuse-skips-end/23504)

Yes, I have these links handy because I've been tracking these issues for years and they're still not fixed.

Thanks for this info. I've been patiently waiting for Thread/Matter devices so that I could just use my AppleTV as the TBR and not need any Zigbee gateway. And not need HomeKit-specific devices; just use generic ones.

But I'm certainly not about to let simple IoT devices have any internet access at all. Being unable to block this on the TBR as you suggested would be mandatory for me, and not possible on AppleTV.

Location: Toronto or SF

Remote: Yes

Willing to relocate: No

Technologies: Cloud Infrastructure, Kubernetes, Docker, Terraform, Linux, Python, CI/CD, Observability, Networking, Distributed Systems, Ethereum, Polkadot, Blockchain Infrastructure, REST APIs, Jira

Résumé/CV: https://jonpurdy.com/static/JP-Resume-2026.pdf

Email: hn202607@jonpurdy.com

Technical Program Manager with 12 years of experience leading complex cross-functional engineering programs from planning through delivery. My background is primarily in infrastructure, platform engineering, developer tools, and blockchain (specifically Ethereum).

I currently split my time between two part-time roles. I advise a small investment firm on technology initiatives, including software development strategy, technical hiring, and execution planning. Separately, I serve as Head of Strategic Initiatives for an affiliated non-profit, leading feasibility assessments and execution planning for major programs.

Both roles are project-based and initiative-driven rather than full-time operational positions. My involvement is around planning, evaluation, and execution milestones, so I'm flexible to pursue a larger full-time opportunity or additional part-time or contract roles.

A few areas I've specialized in:

- Probabilistic forecasting (Monte Carlo), roadmap planning, estimation, and delivery forecasting.

- Building engineering planning and analytics tools (Python/Jira API) for throughput analysis, estimation accuracy, dependency visualization, and forecasting.

- Scaling engineering organizations by improving communication, roadmapping, planning, and execution.

- Technical deep dives with engineering teams to break down complex work, identify risks early, and keep large initiatives on track.

- Developing engineers and technical leaders through clear ownership (DRIs), mentoring, and enabling teams to operate with less coordination overhead.

I didn't, but I'll do so now. I usually don't bother with support, because I forget that good companies like Fastmail actually have a competent support team.

I literally switched on "Enable offline support", caching "All mail" offline on my iPhone a few months ago. Tons of free space, only using 4GB for offline.

But when my phone is actually offline (on a plane or elevator) it beachballs when trying to find something.

Found this a few days ago. It solves a problem I didn't know I had: showing a map of all my photos locally (like Apple Photos), but not copying them or allowing modifications into its own library. I don't want sync, or the ability to edit them, just display and search them (and view and maybe edit metadata).

You just add a folder as a source and it'll create thumbnails, add metadata, do face recognition (if desired, using MacOS APIs), all locally.

One feature I would love: write the name of the recognized person to the metadata as a keyword (which I do manually now in Photo Mechanic).

> Even if they have taken away other routes that used to exist so that this is the only way?

I’ve only been in the system for a year so I don’t know about the history beyond that. Based on some downvotes + your comment I will look into it further.

Very useful comment. I’ve had an A1 Mini for a year now and it has been my favourite purchase in years. Like when I got my first mobile phone, I feel like I’ll have some sort of 3d printer for the rest of my life. Bambu made it super easy and inexpensive for this to happen.

I’m completely against bullying and attempts to lock out open source software from using 3d printers directly; if they locked out OrcaSlicer from direct control I’d have a big problem with that.

But trying to interact directly with Bambu’s private infrastructure/APis seems reasonable for Bambu to block. I think a cease and desist might backfire on Bambu but i don’t think it’s unreadable. (Didn’t watch the video. Just getting context from parent comment. )

Yes, the model ships with Tahoe, not previous versions.

I too would love to try this for simple prompts but won’t be updating past Sequoia for the foreseeable future.

I can think of two instances from the past year or two where this happened: "printer cable" (USB-A to USB-B?), and USB-A extension cable (both at separate times). I think I spent ~$10 for each of these, so my total bill was $20.

So $20 fee to pay for getting rid of a bunch of other cables I didn't need years ago and saving ~500 cubic cm of space.

And I gave the printer cable away to a friend when I was done with it, happy to repurchase it in a few years in the increasingly unlikely scenario that I need it again.

I’m looking for a recommendation to get beyond TinkerCAD (for 3d printing). I learned it in 2019 and came back in 2025 when I got my own printer. It is comfortable and fine for my purposes but lacks basic things like chamfer and fillets.

Anytime I try to jump into Fusion or FreeCAD I immediately hit a wall (like trying pirated Maya when I was a kid).

Totally correct and a good call out. I did check this as best as I could for this particular model of TV. But I'd have to do the same in a few years if it was ever to be replaced. I suspect I'll have to desolder the cellular module of my next TV circa 2036...

I wonder about this every time I see a smart TV-related thread on HN. I recently purchased an LG OLED (C5 48") because my old TV died so I'll finally comment. As others have said, just don't connect it to the internet. But you knew this already, so I'll provide my anecdote on the experience of this since I wondered the same thing for years before getting this TV.

When the TV is never connected to internet, and you use a single HDMI source like me, the TV acts completely like a dumb TV. It gets turned on via my AppleTV remote and displays the picture 1-2 seconds later. No LG logo (I disabled this), and no smart interface shown whatsoever.

If you want to change settings, you can display the settings interface via LG remote control and it generally acts like a dumb TV (not blocking the entire screen, so you can adjust picture quality and see the result as expected).

I've had the TV for about two months and never been asked to update it or shown any ad. The only time I've ever seen the smart fullscreen interface is when you unplug a live HDMI source and the TV detects that nothing is there. (If you turn the source off, it tells the TV to turn itself off as well.)

Hope this helps since it's a lot easier to buy a nice smart TV and do it this way than find a truly dumb commercial panel.

Now that you mention it, DJI action cameras required the same thing (installing an app to activate the camera). I bought one, discovered this during setup, and promptly returned it and got a GoPro instead.

I do something similar with voip.ms (hosted Asterisk).

The intercom calls my voip number, which can be set two ways: 1) play DTMF tone 9 to let the person in, then hang up (which is a security risk if random folks at the intercom buzzed me up trying to get in.

Or 2) plays audio "enter passcode", then:

- if the visitor enters the code that I told them, it plays DTMF 9 to let them in

- if the code is incorrect, plays "incorrect passcode" and hangs up

It also sends an email and SMS whenever someone triggers the intercom so I know about it. With passcodes, I can even set up multiple passcodes to give out to various people (like Amazon, friends) and my notification will display which code was used.

I have the opposite experience (in tech at small or medium companies). Managing remote workers is much easier since outcomes (and outputs) are necessarily more visible.

Before working remotely (pre-2019) when managing teams in person, I found myself necessarily having discussions to get synced with folks. At my most recent role (and previous remote first roles), team members were excellent at providing updates on Github issues (the sources of truth for work items). Of course, this required buy in at all levels and trickling company objectives down through the program(s) and linking work items to OKRs etc. It was very obvious when folks weren't hitting objectives and easy to gather detailed written evidence of this.

And regarding getting to know folks. Most recent offsite was at a villa in Croatia where I got to both meet my team members and ended up getting to know them like friends. Now that I think about it this has happened at previous companies as well during remote offsites.

I wonder if it's field-specific. Sounds like there are multiple anecdotes across a wide distribution of outcomes.

Your description of the test and your replies to questions indicate you've come up with a pretty great assessment for the role(s) you hire for. Especially where you mentioned:

> The test was a few of those questions and a few which were easier to cheat, and almost nobody had good scores on just the cheatable section

I also like how you allow/encourage self-assessment, where if a candidate can't do the test in ~20 minutes under zero pressure, they probably won't be a good fit in the role itself.

This is very useful context. Especially around Contact Scopes etc. It's never made sense to me that iOS shares if the user is choosing to not share their contacts.

Apple seems to basically do privacy-related things to an 80% level but not bothering with getting it totally correct. This makes business sense because the extra 20% is way more difficult, but it's great to see GrapheneOS going all the way.

Glasses would have been the "normal person" fix, but my eyes are great otherwise (better than 20/20 distance vision). So I could focus closer with glasses, but the lenses were worse quality than just sitting farther back.

100% agreed. I'm a Program Manager and have been writing tooling for my own internal workflows for years (like Monte Carlo-based forecasting tools), or program-adjacent low-stakes stuff (like an API to generate a WSJF score based on a fields inputted into Asana, since it couldn't do that itself).

But I'm not about to send a PR for fixing production bugs even if I have decent high level context. Nobody has better context than the devs working on it every day.