persolb

I’ve never had a phantom brake on FSD. I know others have though… not sure what the difference is.

Phantom slowdown doesn’t really happen anymore. I primarily don’t do it because of the stop sign issue and sometime some turn signaling without changing lanes.

So basically… embarrassment. I don’t mind it driving like an old person, but I’m embarrassed to be seen as driving that way.

The latest version is really good, considering. Drives roughly like a bottom 20th-percentile driver.

I use it 100% of the time that nobody is behind me. The worst thing is how it handles stop signs… it actually stops at the stop sign, then rolls forward slowly to where a human would actually stop.

(Note: I got it back when FSD was substantially cheaper; I don’t think it is work the $12k it is today)

The fun part was how broken the script cheat detection was. You could make massive changes, then randomly change a few characters and find a combination that wouldn’t get detected as a change. It ‘felt’ like they were using a 4 bit checksum to detect script changes.

I think the window on this being viable is pretty short. On-PC LLMs can already produce these summaries fairly well. Summarizing results across websites isn’t one of the harder tasks.

The missing piece is a user friendly UI that makes it usable to people who have never heard of GitHub.

Can someone explain what this means? Reproducing some private work publicly?

You probably still have a better perspective than me; at least having worked on these specifically.

I suspect that at least some ships/designs are done right ‘right’.

(1) the engine controllers are internal safety limits and have very controlled input ranges. Cummins engines as an example.

(2) the network has a ‘Battlestar’ mode where you can just cut the wire. People would still need to connect laptops or jumpers locally to control devices in anything beyond a ‘max’ vs ‘idle’ vs ‘off’ mode… but 100%, ready, and off should be enough in an emergency.

I think the last 23 years has shown that, luckily, those people are mostly idiots.

I suspect many people in HN could whip up mass violence with drones if they wanted to. Luckily the people who can generally have better things to do.

Do you have inside industry knowledge here?

I’m in an adjacent industry, with less risk of death or commercial loss, and the compressor backups only output to SCADA. The pressure regulation is all relay based and the on switch is a manual secondary contactor.

I don’t think “don’t be born there” applies to NYC anyway. Less than half of the residents were born in NY state. (So some percent less than that born in NYC itself.)

We need ‘some’ method to allocate residences, and price seems to be the best one we have.

Maybe give people $10k in loans to move out of NYC… they’ll more than save that money by committing from another state.

It isn’t really completely unrelated though. I often drive for 6 hours between cities due to speed, comfort, and cost. I am certainly more likely to die for it.

I think the main response would be ‘then dont live there’. I live in a city cheaper than NYC, but I moved to a cheaper area further out due to cost.

Land is unfortunately limited. Someone else is willing to pay more for the space than I am.

I understand moving sucks and prices in areas change, but this is still an area where a functioning market is better on whole.

Do you think the 1 km wide DMZ isn’t meaningfully changing the situation?

(I obviously don’t like the idea… but from my view there have been multiple attempts to have Gaza develop, and they generally fail out of apparent spite. If the adjacent country is a failed state run by a terrorist group… I’m not sure what better ‘meaningful change’ can be reached.)

The shear volume of death everyday is mid blowing.

I don’t see any benefit to anybody to just ‘stop’ when a small number of those deaths happen in an exceptionally dramatic way.

The repercussions of this accident could very well kill more people indirectly than the actual workers who died.

Honestly a mental swag. To put actual numbers in it, life expectancy was 20 years, now 80 years.

That means roughly 0.9% chance a of death per year now, vs 3.4% ancestrally. Thats 73% of deaths avoided per year. I think that compounds year over year since those people who would otherwise have did are also alive to avoid more deaths.

So, even at the indisputable low end… 73%.

The captain was probably thinking ‘oh shit’. This was almost certainly an accident that the captain wasn’t solely responsible for.

We are already preventing 99% of preventable deaths. As you say, most ‘preventable deaths’ we prevent just turn into a new later more creative preventable death. In many cities you can be homeless and have better mortality prevention than a medieval king.

I don’t know how people think that we’ll draw a line somewhere clear. It is (and should always be) a moving target.

I think most progress has this moving goalpost. Because of this, we need to multitask as a civilization. We can’t just focus on ‘safety is number 1’… because we’ll never do anything else.

I thought this too at the time. I ran the scripts, check my profile to make sure they were gone, then deleted my account. Went back the next day an a lot of my posts were back. And I couldn’t delete them anymore since I killed the account.

But I just checked, and they are gone. I can search google for my old username, and just references to my u/ by others is seen.

You might want to check again; I think there was some cache or mirroring that the deletes took a long time to propagate through.

It almost resembles mind mapping. It is a useful ‘process’ to figure out what you think/know. And it might be a pretty picture. But it isn’t very useful as documentation.

I think your definition of ‘slow’ differs from theirs. An analogy to make it clear:

A self driving car 20 years ago could be in a drag race and do 100mph.

Now self driving goes most everywhere, but at the posted speed limit.

Do people claim it doesn’t?

I think the main reason we don’t see this is that very few places actually ever have ‘mass’ migration, relative to overall population. I found one study, which shows a relatively small effect that is probably hard to see considering all the other things that impacts housing prices: https://www.sciencedirect.com/science/article/pii/S016604622...

Think of it like top down budget va bottom up budget.

RAMS engineers do a bottom up analysis. However, it is fairly typical for the to be subtlety biased in a way that changes the final answer. (Usually by ‘rounding up’ on individual hazards).

Top down takes an outside view and says ‘this looks like these other past projects. It should land in around the same place.’

I’ve found top down tends to be more accurate project wise. Bottom up more accurate for specific subsystems.

Since “I’m in this photo”, I think it is mainly because she ch paths generally don’t seem effective. Changing the structure of society doesn’t happen very often, and tends to have undesired side-impacts.

Seems reasonable to me. I cut the simple bushes in front of my house back from the path much much more often than the township does anything with the road.

Manpower wise, the bushes along the path probably take as much average labor as the road.

Maybe material/gas to resurface the road every 10 years would make the road cost more per year.

It is mostly a matter of time. I have home assistant running locally, but started paying nabu casa to serve as the remote access bridge after my homegrown external site died when I upgraded my internet. I could fix it myself… but it was faster to just subscribe.

The other issue is that some cloud services are just better than what I can do on a raspberry pi. I’d rather get voice recognition from Google for ‘free’ than pay electricity and time for a server to always be on.

I think it’s mostly the ‘everybody agree’ part. There are millions worth of years of effort pit into today’s webpages.

A more likely successful path is to standardize markdown links, and just let browsers start showing markdown natively. Throw in a simple optional style sheet for good measure.

I just want to second this. I search “name” LinkedIn whenever I don’t know who people in a meeting are. It is extremely rare that the LinkedIn doesn’t provide some useful info.

Much easier to answer questions at the right level of detail when you know the audience.

I partially disagree with the first one being a problem. For apps that require ongoing developer resources, I actively seek out subscriptions now. (And some export off ramp)

I’ve been burned too many times by a great app that I have invested time into, only for it to be non-function years down the line after an OS update or a switch to another platform.

This is mostly an issue with knowledge management type software. I’ve moved everything I can to Obsidian, since the off-ramp is just clear text, and still subscribe to it to help incentivize ongoing development.

As an analogy, I like to think of it as us only having access to the compiled/minified/processed code; except it is massive, no libraries were used, and we can’t even fully describe the hardware.

Yeah… technically that code is controlling what happens (along with whatever hardware inputs exist). But the raw HEX is also near impossible to use in a productive way.

There is an Oculus game Tea for God that had step scaling as an option. It is rooms ale, but every step you take, it moves you 1.25 steps of distance.

I get motion sick easy, but this option doesn’t make me sick.