1,023 karma · joined September 5, 2019
Honestly it seems dystopian. Like requiring a breathalyzer before you can turn the key in every car. Treatment like this is supposed to be reserved for those who lost their freedom through egregious actions -- not the entirety of society.
Clearly, the design of the road has an incredible influence over what speeds drivers feel comfortable - and uncomfortable - going.
There is a road near me (SWFL) where everyone routinely goes about 50 MPH, despite a 35MPH limit. The road has three lanes on each side, ditches on either side, a wide median, and dedicated turn lanes to get on and off. It's uncomfortable - and often dangerous - to follow the posted speed.
For any major road, it seems like the best approach would be to build it, then test what speed the majority of traffic goes on it before setting the limit. Get the average speed, add 5, and post the sign. If the speed is too high for the type of road, it should be redesigned. Surely it'd only take a couple of tries at the process before we figured out how to make a road where most people would drive the speed we intend.
--
As for electronic speed monitoring, there is another nearby road that's posted at 50 MPH that google maps insists is still 35 MPH. If electronic speed monitoring goes to its natural extent, that sort of mistake could result in hefty, unjustified fines or even jail time.
In any case, I have a feeling that this is a band-aid solution that will further discourage the average consumer from purchasing new cars.
We were doubly motivated by mosquitoes and the fact that my grandmother - aged 99 at the time, who has since passed - was going to leave without saying goodbye if we weren’t back by the right time (something we weren’t aware of until midway through the hike). Would have been much slower otherwise.
I’ll definitely be using this rule in the future to sanity check the AllTrails estimates.
edit: a more recent hike in Kansas - 20.04 miles with ~1600 feet of elevation gain - I did in 6:45:28, according to my watch. A bit faster than the 7:28 estimate I come up with using the rule.
It came up a lot when making HTML-based forms. I’d define them all in PHP, but the experience is best without a page load, so typically I define some typescript object to send the form via Ajax. I got tired of having to define all the field names in both places (awful for maintenance), so instead typescript uses AJAX to ask the server for the list of inputs it needs to grab and send.
It sounds like Haxe could be a much nicer way to do stuff like that: https://haxe.org/use-cases/web/
I wound up writing a very ad-hoc C++ program that would parse a base SQL file and generate the appropriate DB2 and MariaDB dialect versions. Not flexible enough for reuse but it got the job done.
In security, "covert tests" that replicate rare events (like attempting to sneak a gun through a checkpoint) more frequently, seem to increase the chances of success in detecting real threats [0]. If only we could do the same for our legal system.
[0]: https://www.cmu.edu/news/stories/archives/2013/november/nov4...
It's not the most dangerous thing that man has ever built, but it is dangerous, and it's probably the most uncomfortably personal.
The following was encouraging:
> In late 2021, the scientists began to run new experiments. First, they were curious if an algorithm trained on one person could be used on another. They found that it could not — the decoder’s efficacy depended on many hours of individualized training. Next, they tested whether the decoder could be thrown off simply by refusing to cooperate with it. Instead of focusing on the story that was playing through their headphones while inside the fMRI machine, participants were asked to complete other mental tasks, such as naming random animals, or telling a different story in their head. “Both of those rendered it completely unusable,” Huth said. “We didn’t decode the story they were listening to, and we couldn’t decode anything about what they were thinking either.”
> Neural Trojans embedded in pre-trained neural networks are a harmful attack against the DNN model supply chain. They generate false outputs when certain stealthy triggers appear in the inputs. While data-poisoning attacks have been well studied in the literature, code-poisoning and model-poisoning backdoors only start to attract attention until recently.
While I agree with other commenters here that it's hard to imagine it being widely implemented at the moment, I think it's only the beginning of this type of tech. In a few years, maybe this type of data poisoning could be as ubiquitous as SSL and as subtle as webp compression artifacting.
It happens a lot. I think it's one of the big reasons that hyped-up video games and movies always disappoint.
Opening the hatch? At 13,000 feet below sea level?!
For one, very few people buy physical media anymore - but streaming isn't licensed like physical media is, so while you can buy the disk form of virtually any movie without any hassle, you may have to sign up for a specific subscription service or platform to stream. For example, I would need to sign up for Comcast to stream "Cannonball Run" - but I can buy it on DVD from Amazon.
It also means that anything that isn't available to stream, anywhere, doesn't get watched the way it might if we were still in the Blockbuster era.
And of course, missing bonus features and special cuts of films is yet another major change.
On the whole, I've been able to consume far more content that I love thanks to streaming than I think I've lost because of it. But it's worth noting the changes. Will definitely check out Criterion's service.
I think one of my favorite things about it is actually IBM's Data Description Specification (DDS) [0]. You use it for the same reason you'd use a `CREATE TABLE` statement in SQL, but the syntax is much more legible / suited for that purpose, and the file is stored permanently (and used to spool up a "physical file" with the given format, where the data is actually stored).
Definitely encourage anyone who is curious about these / esoteric OS's in general to look into them - you can get hands on through IBM's "Cloud for Co-Creation and Enablement" (I believe it's been renamed) or PUB400 [1].
[0]: https://www.ibm.com/docs/en/i/7.3?topic=files-describing-usi...
Definitely gave me a leg up on things like spelling, social interaction, learning not to trust strangers who promise to guard you through the wilderness, etc. Arguably guided, subconsciously, some much later intuitions I had about psychology / economics / game theory. And it was probably also responsible in getting me interested in computers.
I could see it being part of a funnel that makes people more likely to wind up on HN.
If I find the dough I really hope to buy one before they're forced to go all EV.
[0]: https://www.penguinrandomhouse.com/books/107178/1491-second-...
However, somehow I doubt that we want to make a database like that given that, even if the hashes are irreversible, the database itself could be used to glean information.
Of course, a better solution could still be the one suggested in the article: stop over-classifying, and reduce access to classified materials overall.
Definitely invest in a good notebook if you get a nice pen. The pelikan is a pain to write with on regular paper.
But if you look, virtually every town of reasonable age has a walkable "downtown" area. Even in small Kansas towns - look at downtown Ottawa, Seneca, Lawrence, etc. Or, across the country - Rock Hill, Greenville, Savannah, Tampa, Orlando, San Diego... The only ones that lack walkable downtown areas are the most recently constructed communities (Nocatee, Florida comes to mind - recent construction, the "town center" is effectively a strip mall).
I started opening GitHub issues I never expect anyone else to read to help myself explore and verbalize a problem without having to nag anyone else about it; it’s been helpful and also yields a lot of what is effectively documentation of the “why” behind our design choices. Not sure if it would work as well on public repos or if you work with anyone that tries to max their GitHub stats (eg average issue close time) though.
Noteworthy pull-quote:
> Most people would see less than half of their workload automated and would probably continue in their jobs, with some of their time freed up for more productive activities.
> In the US, this should apply to 63 per cent of the workforce, they calculated. A further 30 per cent working in physical or outdoor jobs would be unaffected, although their work might be susceptible to other forms of automation.
> But about 7 per cent of US workers are in jobs where at least half of their tasks could be done by generative AI and are vulnerable to replacement.
> Goldman said its research pointed to a similar impact in Europe. At a global level, since manual jobs are a bigger share of employment in the developing world, it estimates about a fifth of work could be done by AI — or about 300mn full-time jobs across big economies.
I am glad someone else feels this way. Maybe it’s not going to be as big a paradigm shift as I originally expected.
We will always want to do the discovery ourselves, and I can see why fighting that instinct is a challenge for those in the field.
I know GPT-4 won’t put us all out of work, but I worry that something could in a number of decades I can count on one hand. Of course, it would be gradual.
I am excited, disappointed, and as a young person who hopes to be a SWE for a long time, somewhat afraid.
Then again, it could be argued that this merely means GPT4 is a compiler…
[0]: https://en.wikipedia.org/wiki/Dunning%E2%80%93Kruger_effect
> The truth is that this trade surplus is not really a choice: it is the outcome of decentralised decisions made by millions of economic actors and in the absence of an adequate mechanism for correction. To put it simply: there is no pilot in the plane, or at least the pilots available are not very accurate.
I do question what my life would look like in that case - for the past 5 years I lived about 7 miles from work. I am currently in a college town while I am finishing up at university, and so I'm now: 3 miles from school, ~40 miles from work.
I use pseudo-public transport (a shuttle that my apartment complex provides / runs privately) to get to school every day. However, that situation works because there is a large group of people that need to go to exactly the same place (campus) at approximately the same time every day. I still need a car, because I often have to go to school late in the evening to use lab equipment when the shuttles don't run.
For work, it's even more important to have a car - a public transport route would likely take ~45 minutes compared to the ~12 minute drive on my old 7 mile commute. For the 40+ mile commute I have now, I'd likely have to jump between several different busses (go from my town to the town my company is at, then a bus from there to the regional airport where my office is). That would probably be at least an hour and a half, each way -- if you could even get the city to run a route from a hub to that regional airport.
All of this is not to mention trips I take just to explore, or to go hiking, or to see friends or family in distant towns, grocery shopping, or visiting a specific specialty store.
From what I've seen so far, the only people whose quality of life wouldn't be decimated by a lack of a car are the most privileged - young elites in major cities who live close to work (and can cycle there), who can afford the fees for an uber or taxi when they need one, or who work entirely remotely. The people in the lower and middle class who commute to work and want to go grab beers with friends a few miles away on their friday night - well, they're either forgotten, or mocked when they're remembered. It leaves a bad taste in the mouth for the whole thing.