HNHacker News
TopNewBestAskShowJobs

lstamour

4,318 karma · joined April 27, 2013

submissionscomments
lstamour··on IKEA-Oriented Development
If you look at steps 5 and 7 in this installation guide https://www.ikea.com/us/en/assembly_instructions/vidga-corne... you'll see that they have a way to tell you to use one screw vs another, but it's also true that Ikea's instructions aren't foolproof. This is just one example of how they really need to continue testing and improving their instructions because clear instructions are hard, and writing instructions without text is just making your job of communicating information that much harder.
lstamour··on HDR QR Code
Technically, then, that display is like OLED TVs pre-2018, which could only display about 70-85% of P3 wide gamut because they could not go bright enough across the entire display. They were amazingly bright if a small part of the panel needed to be lit, though, with excellent contrast ratios. By comparison, most LED panels that are wide gamut but don’t have dimming zones can only display a uniform brightness or dimness and often struggle to make colours pop against darker backdrops the way they do on MacBook screens (with dimming zones) or iPhones (with OLED screens). Point is, just because something is wide gamut doesn’t mean it can display the entire gamut nor does it mean it can display the entire gamut at the same time. Its quality can vary by image and lighting technology (since the quality and brightness of colours all rely on their lighting source, etc.) as well as the image content demanded to be displayed and the lighting of the room (since your eyes have to perceive the screen too, against whatever backdrop or lighting your room has).
lstamour··on HDR QR Code
To clarify slightly - the max brightness is the real feature, and everything else including more bits per channel is just a bonus used to describe colours that are dimmer or brighter in more detail than was possible before. Essentially more bits per channel means greater precision, but the real game changer is how the range has expanded and you can display more details in dim scenes, more details in bright scenes, and, very key for the HDR effect, to have both very bright and very dim content display next to each other. As a side effect, just like how you don’t want your seats to rattle in a movie theatre every time music plays, there’s a difference in visual brightness between “normal” sRGB content and brighter (or dimmer) than usual P3 content. This QR code trick exploits that difference.
lstamour··on Graphic design items from Internet Archive
Technically the screenshot you’ve linked to is the iOS 5 newsstand: https://www.cnet.com/culture/how-to-get-started-with-ios-5s-...

But yes, iBooks did start with the same design: https://www.redmondpie.com/ibooks-3.0-all-set-to-be-announce...

lstamour··on ASRock Brings 56-Core Xeon W3400 to Deep MicroATX
I should clarify, I'm speaking from experience, having built a 9900K system on top of an ASRock mATX board. The water cooling barely fit, and it pumps out heat like no tomorrow. Still, it's undeniably tiny and I irrationally love that build, which I can't say about my previous PC towers. So I'm probably the target market for this...
lstamour··on ASRock Brings 56-Core Xeon W3400 to Deep MicroATX
Sweet. Space heaters keep getting smaller and more functional! ;-)
lstamour··on Trapped Under Trucks
The above is "an emotional, illogical argument". The logical argument is that a burden paid by all is sometimes necessary when there are significant risks. This article goes to many lengths to visualize the risks in examples with and without guards. Admittedly, it doesn't have many counter-examples, nor examples from vehicles other than sedans and older cars. And it cites costs from the 1980s rather than today's costs. But when sedans on the road outnumber trucks, the risk starts to seem equivalent to a flight safety issue where we've known for years that the planes caused fatal injuries but we never addressed it.

If there's a blindness in this article, it's not so much the trucking industry response as it is the question of whether or not this has been implemented anywhere else in the world? https://www.reifflawfirm.com/canadian-and-european-truck-und... suggests that, "in Europe since 1994, front underride guards and side underride guards have been required on large trucks," and that these were strengthened in 2007.

That same page suggests front and side (of truck) collisions with a passenger vehicle are more likely than rear collisions, and that existing US and Canadian protections aren't strong enough.

lstamour··on What Reddit Got Wrong
Highlights how hands off Reddit generally was until now that while we say “Twitter” vs “on Twitter” or “YouTube” vs “YouTuber”, until now we rarely distinguished between “Reddit” and “on Reddit”. Reddit was an invisible benefactor until it wasn’t…
lstamour··on Intel is all-in on backside power delivery
Adblock Plus like many adblockers gets industry funding for a whitelist of “acceptable ads”. uBlock Origin, 1Blocker, NextDNS, ControlD, and sometimes AdGuard tend to block more. The DNS adblock options are particularly interesting as they can block ads in any app, similar to running PiHole or a hosts file, but in the cloud. There are also extensions to change the website to dark mode (Dark Reader), accept some but not all cookie prompts on your behalf (Consent-O-Matic) and block “Open in App” messages (Banish) or annoyances (Hush) or other annoying behaviours (StopTheMadness). Basically… you can have a browsing experience that’s nice and sane, and just the way you want it.
lstamour··on First impressions: Yes, Apple Vision Pro works and yes, it’s good
I think the problem is the plural. You could convincingly say that 24 is _an_ order of magnitude bigger than 4, even if technically log(24) is 1.38 … it’s a close enough approximation.
lstamour··on Dennis Ritchie’s page on Bell Labs website
The real problem is viewport meta and how you have to opt in or opt out: https://developer.mozilla.org/en-US/docs/Web/HTML/Viewport_m...

If you opt in, a simple page like this ends up looking pretty modern anyway, but plain, while if you opt out, your site is often rendered as if made for an older desktop monitor, like this one.

Another way of putting the problem, we preserved old desktop styles rather than invent new default responsive styles for touch-based browsers.

Without a standard like grey backgrounds and blue links to fall back on, websites had to start making decisions, especially when sites go responsive. At that point it is easier to invent new styles… there aren’t that many useful defaults to fall back on.

This is particularly true for HTML5 form controls, which rarely look good in any browser without styling…

But perhaps there’s hope for the future as new design language becomes HTML tags. E.g. details has an arrow, datalist builds a native autocomplete, and so on. It’s a shame that dialog doesn’t have native formatting tho, or that the idea of creating new, modern default styles hasn’t caught on yet in a cross-browser way.

lstamour··on Witch – macOS window switcher replacement
Not always… sometimes the app is just Spotify. Haven’t tried the option trick with it, but there the window is simply closed. Clicking the icon in the dock opens a new window, but tabbing to it does not. And yeah, I never remember the shortcut to open it once I switch to it, because cmd+0 and cmd+1 don’t bring up a new window for Spotify. So I always reach for the mouse and click the icon in the dock.

Normally my favourite method to switch apps is Spotlight, accessible with ctrl+space or cmd+space, I forget which. Anyway, Spotlight is a great app switcher, you can disable most everything else and just type a letter or two to switch. I use Powertoys on Windows to get a similar switcher there.

But again, Spotify breaks this, and switching using Spotlight does not recreate the Spotify window, if I remember correctly. Once again reaching for the dock…

lstamour··on A medieval comedy act has been discovered in first-ever find, researcher says
Interesting. The table of contents is rewritten in modern cursive, though I can't read everything even there.

  Contents
  The Hunttinge of the Hare.
  A Mock Sermon.
  ...
From there we can flip forward to the actual content and ... yep, can't read it. Honestly the only reason I know it says "huntinge of the hare" centred at the top is because I read it in the table of contents.

I can tell it's English, but that's about all I can tell, and that only because I can read some of the words, but not all of them. :) Maybe I can send the image to GPT-4 for transcription and translation. (lol)

lstamour··on Plane: Open-Source Alternative to Jira
Edit: I should clarify, supporting open source is the best possible model for price pressure and innovation, especially in the cloud hosting space, but I also wanted to let other non-profits know about the discount if Jira is too expensive per-user due to volunteers, etc.

FYI, Atlassian has a non-profits license. While the cloud version only has a discount so you still have to pay per user, the on-prem version is free. Best part about running on-prem under this license is that a lot of on-prem extensions have opted in to atlassian marketplace licensing and offer free plans for non-profits. While this would be expensive to do in the cloud, because the plugins run on your server, it’s basically free for everyone involved. The downside is you still have to manage and renew licenses, and don’t expect too much support given the free price.

> Atlassian also offers free Data Center (self-managed) licenses for registered charitable non-profit organizations that are non-government, non-academic, non-commercial in nature, non-political, and have no religious affiliation.

More at https://www.atlassian.com/licensing/purchase-licensing#trial... under the faq question about non-profit pricing…

lstamour··on Apple announces multibillion deal with Broadcom to make components in the USA
I dunno, I think it’s actually Pick three:

  - extreme corporate profits
  - better wages
  - US jobs
  - lower prices
You can’t optimize both for extreme shareholder returns and extreme wealth at the top and also pay first world good wages/living standards and have lower prices. Unfortunately, in this race, shareholder returns is often still the last thing compromised on, it seems.
lstamour··on Noise is all around us
Re trucks beeping - that’s partly legal responsibility isn’t it? I mean, we have had the ability for trucks to beep as they back up for far longer than we have had the ability for them to realistically have functioning rear view cameras. And since the normal expectation is that vehicles go forward, it makes sense to alert when doing something unexpected, like backing up. And so the laws we have mandate and expect beeping noises from large vehicles because the idea that laws would mandate rear cameras hasn’t caught on yet.
lstamour··on ChatGPT app for iOS
Legal, translation and limiting demand to a reasonable level with rolling release to other markets in future?

That said, I also find it somewhat unfair, as likely does anyone not in the US :)

lstamour··on ChatGPT app for iOS
Yes they work but you will be prompted to sign in with the account that bought the app by entering its password. Auto-updates sometimes work but only if you’ve entered the password for the account recently. Some work was done around iOS … 14 maybe? to try and improve this experience, but iOS is still not fully multi-account. And strange things sometimes happen to DRM content, in-app purchases and music libraries.
lstamour··on Heather Armstrong has died
Yeah, that was my thought too. It felt far too direct and oddly intimate in what it cited and quoted. I asked ChatGPT to write a generic obituary in an NYT style and it came back with things like, “As news of [Name]'s untimely death spread, fans and colleagues flooded social media with tributes and messages of condolences. Many praised [Name]'s kindness, generosity, and unique voice, and expressed their shock and sadness at the loss of such a talented and beloved figure.” And “[Name] will be deeply missed by their family, friends, and countless fans, who will always remember them as a bright and inspiring presence in the online community.” Which feels like the right way to approach the topic. This, on the other hand… focuses far too much on written word and research instead of impact and legacy. It felt tone deaf.

For comparison, here’s an article from a fan: https://twitter.com/ejdickson/status/1656425063321677825 - or for the direct link, https://www.rollingstone.com/culture/culture-features/heathe... Not that the article is any less frank about it, but at least it addresses some other points.

And it’s not like there aren’t other posts from contemporaries to quote. There are, just search for them: https://twitter.com/anildash/status/1656344993995411456 for example.

This is more what I expected to read: https://www.washingtonpost.com/opinions/2023/05/10/heather-a... or https://www.washingtonpost.com/technology/2023/05/10/heather...

lstamour··on Google keeps failing its Nest customers
So there are other things going on here and I did end up moving from Nest to Ecobee and mostly am okay with the switch. (The bar is very, very low right now.)

But I wanted to clarify about video recording and viewing 100% of clips:

there are two service levels from Google: either you’re grandfathered on older pricing plans per camera and can see a 24/7 recording or you upgrade to new plans per household that they sell now where the camera only records and sends motion or sounds, not 24/7 feed recordings like the old Dropcam used to.

You can’t buy the old way of recording 24/7 from every camera as far as I know, and the only reason I can think of that they switched recording modes is because it saves power and is more efficient so hardware might not break as often via overheating. I mean, they might claim it is more efficient for storage, but they have stationary cameras - compression has to be good, right?

Anyway, just pointing out that generating clips from video recordings is kind of the old way, and the new way is that it somehow automatically makes clips for you as motion alerts.

But… outside of that I agree 100% that Nest has been basically dead since the rebranding. That Ecobee still has any unique features Google doesn’t have largely stems from this lack of innovation.

Ecobee is now owned by a company that makes generators, the kind that runs on gas or diesel. It’s hard to imagine that Google can’t ship new features, apps or software products faster than a company that makes industrial appliances and yet here we are.

Google makes too much money from ads to care about any other way to make money, long term. I’m shocked they don’t just give away these smart devices and wire them up to play YouTube ads with facial recognition at this point.

Maybe they’re waiting for Amazon to try it first, or want to experiment with Google TV ads like Fire TV and Samsung already do. If my Apple TV ever starts auto-playing trailers and ads, I’m going to permanently switch to using AirPlay, I think.

As for the Dropcam, maybe I’ll replace it with an iCloud video camera when the time comes…

lstamour··on Should You Give Your Kid Melatonin?
Correction for the article: melatonin is available without a prescription in Canada (or at least Ontario) and in France. I know from personal experience. In the UK, however, I was surprised to see it required a prescription. I can’t speak for the other territories on the list, but I think the whole article might need a fact-check as a result. :)
lstamour··on Google “We have no moat, and neither does OpenAI”
Sort of. The problem with most integrated GPUs is that they don’t have as many dedicated processing cores and the RAM, shared with the system, is often slower than on dedicated graphics cards. Also… with the exception of system on a chip designs, traditional integrated graphics reserved a chunk of memory for graphics use and still had to copy to/from it. I believe with newer system-on-a-chip designs we’ve seen graphics APIs e.g. on macOS that can work with data in a zero-copy fashion. But the trade off between fewer, larger system integrated graphics cores vs the many hundreds or thousands or tens of thousands of graphics cores, well, lots of cores tends to scale better than fewer. So there’s a limit to how far two dozen beefy cores can take you vs tens of thousands of dedicated tiny gfx cores.

The theoretical best approach would be to integrate lots of GPU cores on the motherboard alongside very fast memory/storage combos such as Octane, but reality is very different because we also want portable, replaceable parts and need to worry about silly things like cooling trade offs between placing things closer for data efficiency vs keeping things spaced apart enough so the metal doesn’t melt from the power demands in such a small space. And whenever someone says “this is the best graphics card,” someone inevitably comes up with a newer arrangement of transistors that is even faster.

lstamour··on Google “We have no moat, and neither does OpenAI”
It’s like a database, I imagine - so the answer is probably “unlikely,” that you need memory per-request but instead that you run out of cores to handle requests?

You need to load the data so the graphics cards - where the compute is - can use it to answer queries. But you don’t need a separate copy of the data for each GPU core, and though slower, cards can share RAM. And yet even with parallel cores, your server can only answer or process so many queries at a time before it runs out of compute resources. Each query isn’t instant either given how the GPT4 answers stream in real-time yet still take a minute or so. Plus the way the cores work, it likely takes more than one core to answer a given question, likely hundreds of cores computing probabilities in parallel or something.

I don’t actually know any of the details myself, but I did do some CUDA programming back in the day. The expensive part is often because the GPU doesn’t share memory with the CPU, and to get any value at all from the GPU to process data at speed you have to transfer all the data to GPU RAM before doing anything with the GPU cores…

Things probably change quite a bit with a system on a chip design, where memory and CPU/GPU cores are closer, of course. The slow part for basic replacement of CPU with GPU always seemed to be transferring data to the GPU, hence why some have suggested the GPU be embedded directly on the motherboard, replacing it, and just put the CPU and USB on the graphics card directly.

Come to think of it, an easier answer is how much work can you do in parallel on your laptop before you need another computer to scale the workload? It’s probably like that. It’s likely that requests take different amounts of computation - some words might be easier to compute than others, maybe data is local and faster to access or the probability is 100% or something. I bet it’s been easier to use the cloud to toss more machines at the problem than to work on how it might scale more efficiently too.

lstamour··on NYPD urges citizens to buy AirTags to fight surge in car thefts
Edit: I’m debating deleting this post. It doesn’t have any references to back it up, and I just realized that if the tracking network encrypts the location with the public key such that the private key is required to decrypt the location, it is indeed possible for Apple to not know where devices are because the decryption of the location might be done on your device rather than on Apple’s servers, if they wanted to.

Original post follows: I haven’t read any white papers on how this works, but I’m going to call “slight BS” anyway. Technically, Apple knows the location of the AirTag, that’s how their network works. The payload of airtag data might be encrypted, but Apple also keeps the keys for your AirTags so that you don’t have to associate new AirTags with each Apple device you want to use Find My on. While I don’t doubt that Apple strictly reduces what information they log or identify about AirTags to literally the bare minimum that keeps the service running, it’s a bit ingenuous to suggest Apple doesn’t know. What I mean by this is Apple isn’t lying when they say they don’t know the location of your AirTag (until you use their app to look for it) but it is true they know the location of all AirTags synced to their network, they just don’t know which one is yours until you go looking for it. Encryption probably plays a role in that asymmetric encryption would allow Apple to not know any AirTag details until your devices provide the decryption key. A handshake could be involved too so that your device can’t decrypt the location of the tag without proving they own the associated encryption key. But presumably Apple has a way of issuing or remotely authorizing against such keys. It is likely the case that your password for your phone protects Apple from gaining access to the keys thus justifying the position that Apple doesn’t have access because they don’t have your keys. Apple would have to act maliciously with their software into order to break this system, and it’s not worth it for them to do so. I’m not saying they’re lying but they are simplifying things a bit.

lstamour··on NYPD urges citizens to buy AirTags to fight surge in car thefts
Actually re non-iPhone detection: I’ve been annoyed by air tags beeping and it’s definitely the case that the new anti-stalking features on AirTags make them rather noisy and easy to find as a result. You have to get very creative with the hiding spot so that thieves know it exists (they can hear it) but they can’t find it, remove it or take out the battery to disable it. I’ve heard stories about people planting more than one AirTag on their bike, such as dropping one in the frame in a hard to reach area, and having opportunists decide to drop the bike off at a used store than deal with the AirTag they can’t find that keeps making noise.
lstamour··on Tell HN: iOS converts units for highlighted text
Weird. I’m in Canada and I see the following when I select “1 cup”:

0.24 L | 23.66 cL | 236.59 mL | 236.59 cm^3 | 0.0625 gal | 0.5 pt | 0.25 qt | 8 fl oz | 16 tbsp | 48 tsp | 0.00836 ft^3 | 14.44 in^3

My “measurement system” under language and region in settings is set to “Metric”. Perhaps yours is incorrectly set to “UK”?

I’m also running iOS 16.5 beta, if it helps.

lstamour··on Why is OAuth still hard in 2023?
Not exactly. With webauthn the complexity lies in frontend - you have to explain how and why users should use this and what it is, how to manage enrolments and so on. Touch ID, Windows Hello and Passkeys might simplify this but it is still more complex to explain to less technical customers than, say, Face ID on an iPhone app to remember your login.

If you want an easy all-purpose login flow, sign in by clicking a link in email is still an easy fallback.

OAuth from multiple providers also has complexity - Auth0 makes it so you have to maintain separate databases for passwords if you want to support login via OAuth and login from password. You have to link accounts to login from multiple providers.

Logins are simply hard work no matter how you slice it. Eventually they will be easy but… I’m not holding my breath. :)

lstamour··on SpaceX, Rogers to connect mobiles phones to satellites in remote Canadian areas
If you go back to when T-mobile was incredibly competitive, plans were often $20-30. If I want a month, I can get a SIM prepaid for $30 or less with all the data in the US, and sometimes as little as $3 in Europe.

Plus there was a time, albeit almost a decade ago now when the dollar was on par. Even about a year ago the dollar was still aiming for 0.8 rather than the 0.74 it’s flirting with right now.

Lastly, if you want an Apple Watch with cellular, you end up needing a more expensive plan from the big three rather than their sub-brands. And the Canadian government is threatening regulation to encourage carriers to offer more data for less. There was a time within the decade when $60 at Fido would get you 2GB and that was during a “double data” promo. This was 2018. https://forums.redflagdeals.com/fido-60-unlimited-canada-wid...

lstamour··on Prompt engineering vs. blind prompting
Maybe a difference between prompts or v3.5 vs v4. I asked ChatGPT v4 and it gave an answer that used capnproto::serialize_packed::read_message. I asked it about the difference between read_message and deserialize_message and this is the response it gave (truncated, as it got some parts wrong):

> In summary, you should use read_message when working with packed messages and deserialize_message when working with unpacked messages. Make sure you choose the appropriate serialization and deserialization functions based on the format in which your messages are stored or transmitted.

Google was surprisingly little help on the topic. At best it pointed me to https://capnproto.org/encoding.html#packing which covers the general idea but glosses over parts.

The problem with ChatGPT is you can never be certain or confident that its answers are correct. It’s as useful as rolling dice in guessing a number sometimes. With the training from the internet, the dice are loaded, but the answer is still likely to be wrong because it just assembles words together.

lstamour··on Prompt engineering vs. blind prompting
If you don't mention a library by name it is liable to make something up by picking a popular library in another language and converting the syntax to the language you asked for.

If you ask for something impossible in a library it will also frequently make up functions or application settings. If you ask for something obscure but hard to do, it might reply that it's impossible but it is possible if you know how and teach it.

I sort of compare prompt engineering to Googling - you sometimes have to search for exactly the right terms that you want to appear in the result in order to get the answer you're looking for. It's just that the flexibility of ChatGPT in writing a direct response sometimes means it will completely make up an answer.

There's also a limitation that the web interface doesn't actually let you upload files and has a length limit for inputs. For Copilot, I'm looking forward to Copilot X: https://www.youtube.com/watch?v=3surPGP7_4o

← PreviousPage 6 of 34Next →