The difference is that taxis would either give you a call that they were here or they'd just wait for. They don't care either way if you show up or not because the meter is running. The Uber is gone if you don't show up in 5 minutes. That is if you are lucky and the driver didn't mark themselves as "here" when they were 2 blocks away, which seems to be the norm here now.
Exactly. Sure, I can go read a book and learn about Rails. If I wanted to be a Rails specialist I probably would, but I don't want to be a Rails specialist. And like you said, other languages can do this too, but I haven't really found it to be too much of an issue. I remember one time where I had to break out whatever the rustc flag is that just produces the code with expanded macros, but that's it.
I don't think I've ever been unhappier than working on a Rails project. See a bug on the site - something is rendering incorrectly. grep to find the view. Great, there's a method that is being called to render the buggy section. Grep for the method name - 0 hits. Amazing, it's something that is synthesized somewhere and I have no idea where. Guess I'll stop what I'm doing and read docs for an hour. If you do nothing but use Rails all day, sure, but the whole convention over configuration thing is such a huge anti pattern to me.
Unfortunately you see quite a few manufacturers implement Matter in a way where they require a proprietary hub, and the proprietary hub requires a constant internet connection to work...
I've worked at Amazon since 2018 and they've always talked about the software engineers being fungible during my tenure there. Technically everyone is supposed to be able to do everything, but in practice it's a huge headache if you want to hire for a more specialized role. They started creating some, like Frontend Engineer and Embedded Systems Engineer, but in practice these are still extremely broad
Apple App Store Connect has this issue for example. If you remove or downgrade someone from the store they retain access for however long... and they keep the ability to do destructive actions like remove or change the permissions of other users, at least they used to until a year or so ago when I last checked. They closed my security report as "by design" btw.
To be fair here if you look at their "supporting evidence" most of it is pointing at markdown formatting. Some of the other things it points to are also just common in human writing, though the overuse leans more towards coming from an LLM, e.g. two instances of groups of three back to back with inconsistent use of the Oxford comma:
> If you're building a web app, a mobile app, or a first-party API: JWT is the wrong default and you should stop reaching for it. A row in Postgres with a bearer token in front of it is faster, simpler and strictly more secure.
I skimmed the study rather than the article about it and I don't see them define it at all. They just had a machine learning model take accelerometer data and classified it into "sleep, sedentary behaviour, light physical activity and MVPA". Whether any form of walking counts as light or moderate in this classification is really anyone's guess
I played around with local LLMs on my M4 Max 64GB this weekend and this is exactly what I found. I put Opus 4.7 "head to head" on the same task as Qwen 3.6 and a few other local models. The 35B did not perform well IME - it needed a lot of handholding and even then the final result did not work until a few more tweaks, while Claude one shot the task. The 27B was much better and also one shot the task, but took about ~55min as opposed to about ~15min for Claude. The 27B is probably something that I could happily run for many use cases if I had some faster hardware... the main problem there seems to be that at larger context sizes, prompt decoding can take several minutes.
IME there is a large difference in quality in what is available at the super market. Sure I can do a once a year road trip to Monterey. The average organic heirloom tomato at Whole Foods or Trader Joe's is worse than the average organic heirloom tomato at Spar
libghostty makes it pretty easy to do. I spent about two weeks setting something up until it was advanced enough to daily drive. I wanted to have a modal workflow similar to vim or tmux copy mode, but without having the overhead of using tmux... that's probably a lot more complicated than "I want Ghostty but with $X tweak". You can poke around in the repo to get a feel for what's involved if you want: https://github.com/milch/mistty
I live in Seattle and anecdotally I have seen the number of people running red lights absolutely explode in the last two years. Literally from seeing once or twice pre-COVID to at least one a day. This is not an exaggeration, there's a particular light on my commute that I see at least one driver run per day. My theory is that in an effort to make the intersection safer they adjusted the lights so now there's a period where cars all have a red light while pedestrians are crossing. Meanwhile a certain segment of the population sees all cars in the intersection stopped and decides to slam it. It's a recipe for disaster given there's a middle school down the road from that light...
The poster I replied to made an over exaggerated statement about the prominence of these two words, which I found hilarious, in the same way a teenager saying "I am LITERALLY DYING right now" after the barista spelled their name wrong on their coffee order would be hilarious. This was a very slight inconvenience to the poster's day (at best) that they could have dropped after deciding they don't care about events happening in a country they don't live in.
It's kind of hilarious to call 2 single words at the end of a blogpost that you could have read over "pushing politics down your throat". As you said, you didn't even have an idea what it meant. For all you knew before you looked it up, he was complaining about the frigid winter weather.
I'm not sure where you're getting that from in my comment. I never said US citizens should want H1Bs for everyone with zero vetting, only that they are a net tax positive.
It's not a dichotomy of maintaining the status quo or getting rid of H1b completely. At least in big tech companies, they do follow labor market tests and prevailing wage tests and so on that are designed to vet that there is an unmet need and that visa holders aren't underpaid. I won't deny there are visa mills and consultancies that game the system and pretty much explicitly just hire cheap foreign labor, but this is a thread about H1B in the context of Amazon layoffs, not InfoSys layoffs.
If you have access to data that shows big tech is preferentially hiring visa holders over US citizens you should get on that class action lawsuit right away. That's probably hundreds of thousands or even millions per person in lost wages, and even after lawyers take their 30% cut, that's still a sizable chunk.
Immigrants pay social security taxes, unemployment taxes, ... that they also will never be able to benefit from. Those are purely for the benefit of US citizens
No, because H1B has pay requirements. As someone who went through the process with Amazon I can confirm that they definitely do offer you a salary that is in line with the local market. There might be lower incentive for raises down the line, but that's a conspiracy theory at best
I don't think they necessarily compete for the same market as some of these other routers. This seems way more compact than many of the other options on the market. I just briefly looked around on Amazon and even many other wifi 5 routers look to be about 2x or thicker than this one. Compared to the GL.inet Opal for example, it's about 20mm smaller in each dimension: 118 x 85 x 30mm (Opal) vs. 95.95 x 65 x 12.5 mm (Unifi). The Unifi is pretty close to a tiny 5000 mAh portable battery.
Now what I'd be really more interested in a Pro version, more so than wifi 6, would be a built-in modem with SIM/eSIM.
I find for me that doesn't work. I have to actively count calories to lose or maintain a healthy BMI range. If I don't, no matter what I eat, I end up stabilizing in the overweight category. I'm lucky enough that my wife has a very flexible work schedule where she is able to cook most meals for us, but she hates weighing everything out so I can track... we've tried to stop the tracking multiple times but it never works. I'll think I'm eating to the same satiety level as before, but I end up slowly creeping back up in weight. My wife is fine either with or without tracking for maintaining her weight, though she does also seem to have an easier time losing when she's counting her calories. I think this is why it's so hard to have discussions about it, people's experiences literally don't match even when they try the same things. I suspect I'd be one of those people that if they tried Ozempic would be able to lose tons of weight, and who would shoot back up as soon as they stopped, unless I track the whole time... at which point I wouldn't need the Ozempic in the first place
Trains are also just more comfortable. More space, more comfortable seats, more space for luggage, you can walk around, better bathrooms, easier to work from especially in the 4 seat configuration, … Personally I would always prefer the train even if it is a bit slower. Once you account for traffic a bus that is scheduled to be faster ends up slower anyway, especially when you really needed it to be on time
My impression is purely from a metrics perspective, people who were underperforming can really look like they are 3x as productive. AI is a real increase for them because it can do things they couldn't. It just gives them the ability to publish a thousand lines worth of PRs a day, which the regular and over performers have to review, but that shows up in THEIR metrics, not in the underperformer's metrics. If all you look at is metrics and KPIs and have no technical understanding, this looks amazing to you.
Most people I've worked with that were already some of the most productive before AI took off are still at the top, and AI didn't move the needle much for them. There's simply no way for them to do 3x the work.
I use Kagi at home and Google at work, and IME Kagi results are much better compared to Google. It's easy enough to jump to Google by adding !g, but I never find myself actually using that... I do however end up doing the reverse, where I'll not find the result I'm looking for on my work computer and then quickly repeat the search on Kagi on my phone. I used to use DuckDuckGo on my personal devices before, and while it was mostly fine, I did have to jump to Google more frequently. In the end sometimes the results were better on DDG and sometimes on Google. I haven't really found any situation where Google results were better than Kagi (I do tend to prefer Google Images results, but I only very rarely use image search, maybe once a month).
There was also an instance recently where I was helping out a coworker get set up with something. I had done it earlier in the day, so I told him something like "type in $SEARCH_TERM, then go to the first page and copy the commands from there". He put it into Google, couldn't find the right result, tried a few variations, still couldn't find it. I pulled Kagi up on my phone, searched the same term, and slacked him the page while he was still looking around on Google.
Some people are contrarian simply to be contrarian. I'm loving the features that are coming out. I use Translate, Assistant, Universal Summarizer, and (of course) Search multiple times a day. Not everything is going to be for everyone, I'm personally not really interested in News, but it certainly feels like I'm getting value out of my subscription. The only thing I'm actively missing with Assistant is a proper app - the PWA is fine but the keyboard glitches out sometimes, navigation is not as fluid/smooth as it could be on a native app, same with file/photo uploads, and notifications when you leave the app and the reply is done would really tie it together.
I've already used the Research assistant half a dozen times today and am super happy with the outcomes. It does seem to be more trigger happy with doing multiple searches based on information it found in earlier results, and I've found the resulting output to be reasonably accurate. Some models in particular seem to never want to do more than one search, and you can tell the output in those cases is often not very useful if the sources partially contradict each other or don't provide enough detail. The best I've found to avoid this is o3 pro, but o3 pro is very slow and expensive. If the Research assistant gets 85% of the results in half the time as o3 pro...
Depending on the CLI tool you could even forego memory management completely and just rely on the OS to clean up. If your program completely reads arbitrary files into memory it's probably not the best idea, but otherwise it can be a valid option. This is likely at least partly what happens when you run a benchmark like this - the C++ one cleans everything up nicely if you use smart pointers or manual memory management, while the Java tool doesn't even get to run GC at all, or if it does it only cleans up a percentage of the objects instead of all of them.
I shudder to think what extra features they cram into Outlook for Windows considering Outlook for macOS is already bloated to hell. I've had to use it for the past few months due to our company upgrading the internal AD version (or something) while they built out the support for native macOS Calendar/Mail, and I hated every second of it