HNHacker News
TopNewBestAskShowJobs

xenadu02

10,977 karma · joined May 14, 2014

all comments are my own personal opinions
submissionscomments
xenadu02··on Find SF parking cops
Fun fact: as a regular citizen you can apply for such a permit online. I did this when we were moving houses to ensure the entire block in front of our house would be clear of cars (or we could tow if required) to make it possible to get a moving truck in on the one-lane street where double-parking would block all access. I filled out the forms. A week or so later the city dropped signs on the sidewalk warning of the upcoming closure.

You can also close your entire street for a block party. You just need a certain number of people on your block to sign the form approving it.

xenadu02··on Find SF parking cops
FWIW the sweeper comes by my street on the posted schedule. Most days one or two parking cops come ahead of the sweeper writing tickets. I have never seen them come behind the sweeper though I have seen the sweeper wait for them. I believe it is policy not to write tickets after because as other posts have noted it is perfectly legal to park right after the sweeper comes through even if it is still sweeper hours.

Of course we are on the corner and the other street does not get sweeping (it is also concrete). I assume that is because it is too steep.

xenadu02··on US backpedals as Hyundai factory ICE raid enrages South Korea
Even if there was some technical immigration violation here its not like any of the people involved were some sort of flight risk or criminal masterminds.

They could have simply notified the company and the people about a potential problem with their visas/paperwork and asked them to provide documentation otherwise or if their visas were expired/improper to return to Korea and apply for a new visa.

xenadu02··on UTF-8 history (2003)
It is worth reading the history of the proposal. The final form is superior to the others so someone was doing a lot of editing!

Take the final and second form where the use of multiple letters was eliminated, instead using "v" to indicate bits of the encoded character.

I also chuckle at the initial implementation's note about the desire to delete support for 4/5/6 byte versions. Someone was still laboring under the UCS/UTF-16 delusion that 16-bits was sufficient.

xenadu02··on How big a solar battery do I need to store all my home's electricity?
It's the best kind of revolution: one that doesn't need permission from gatekeepers. At least not anymore.

Don't underestimate the value of decentralization in some scenarios.

xenadu02··on BCacheFS is being disabled in the openSUSE kernels 6.17+
So FSKit is public API and supports userspace filesystems but on macOS most built-in filesystems are kernel modules. On iOS everything except APFS is userspace.

Modern Darwin is mostly a monolithic kernel, though more and more things are moving to userspace where possible (eg DriverKit).

One interesting side-effect of various spectre mitigations is silicon improving the performance of context switches. That has the side-effect of decreasing the cost of kernel/userspace transitions. It isn't nearly as expensive as people still believe - though it isn't free either.

xenadu02··on TikTok has turned culture into a feedback loop of impulse and machine learning
I deleted TikTok and my Twitter account. Also stopped watching TV. I'm happier and more productive.
xenadu02··on Tiny microbe challenges the definition of cellular life
> Take away the mitochondria [...] can cells live on their own?

Neither can live without the other. Too much genetic exchange has taken place in some distant ancestor where critical genes were deleted from mitochondria and moved to the host. Meanwhile host cells became utterly dependent on mitochondria for energy production. Or you might say: the mitochondria were producing so much excess ATP the host cells started evolving to depend on that much energy being available.

The exceptions are later cases (like a few organisms that have copied energy production from the mitochondria genome then later lost the mitochondria entirely).

For all purposes mitochondria are zombie archaea (not bacteria). Hollowed out empty shells retaining just enough function to perform aerobic respiration and reproduce. There is little pressure to evolve away from this local maxima. What benefit would the host cells derive from getting rid of the mitochondria? Not much. And having those critical functions isolated in what amounts to a pseudo-organelle with its own DNA protects it from a lot of sources of damage/error.

So... are we obligate symbiotes? Or have mitochondria hyper-evolved to such a point they are just organelles in our cells - just ones that carry their own DNA instead of relying on the cell's main DNA? Like much of biology... a bit of both in a fuzzy mix without a clear line.

xenadu02··on Pixel 10 Phones
> It's smaller, so it should cost less

That is... not how the physical world works. The laws of physics hate the large and the small. Or perhaps less glibly parameters do not scale equally. Making a phone is more difficult and expensive than making a laptop for the same reason a 30ft tall human would break their own legs attempting to walk.

To put it another way: A thread rolling screw machine can churn out 12mm/0.5" bolts all day long for a penny each. But if you want to make tiny screws for small pocket watches you're going to pay more (relatively) even though that tiny screw contains way less metal than the larger bolt and the operation is similar. A .00001" error in the larger bolt threads doesn't matter. That much error makes the tiny screw completely unusable. Making a thread-forming die with less than .00001" error is very difficult and expensive and the one for smaller screws accumulates error faster relative to allowable error so must be replaced more often. The steel is just as hard in both bolts but the form of the tiny one is proportionally much thinner.

And similarly if you want a 6m/12ft long bolt you are going to pay a lot more than just the proportional cost of the extra metal because finding machines that can even put that much tool pressure on the dies is not easy. It has to be lifted with a crane. It is just more difficult in every way.

Miniaturization is more expensive. Water and dust proofing is more expensive.

For most things there is a sweet range where cost is lowest and utility highest. Prices go up on either end of that middle ground.

xenadu02··on Pixel 10 Phones
The number of people who aren't vocal tech people who actually want a smaller phone is a very small part of the market. In HN-like circles they're a notable minority but among the general population they are a smaller percentage. Especially when you consider huge segments of the market where your phone is your only computing device: a smaller phone is a massive anti-feature in large parts of the world.

Plus almost everyone who says they want a smaller phone will just buy a larger one anyway.

The sales numbers just don't justify it. Like people who pine for manual transmissions: they're vocal in car forums and publications but they're a tiny minority and making one is a money-loser even in the sports car segment.

xenadu02··on Ask HN: Why does the US Visa application website do a port-scan of my network?
Gaming of the procurement system. The websites are all written by big consulting outfits. Not to mention the disaster that is big corporate IT projects combined with government rules.

Obama had the Digital Service (that Trump shut down) which paid higher salaries. Those folks were sharp and everything they touched was actually decent.

As I noted this is not unique to government. Large corporate projects at the Fortune 500 are often the same sort of consultant-driven crap.

xenadu02··on Someone keeps stealing, flying, fixing and returning this man's 1958 Cessna
Since I can't edit:

I mean "Class B" airspace, which is mostly used in the USA. The ICAO definitions specify what the airspace classifications mean generally but each country often has its own slight variations of the rules.

Class B/C/D is airspace where contact with ATC is required but differs in separation policies. IFR traffic is always kept separated from other IFR traffic.

A: IFR only B: All aircraft kept separated by ATC. IFR-VFR and VFR-VFR are all ATC's responsibility. C: IFR traffic kept away from VFR traffic. VFR must watch out for other VFR. D: IFR must watch out for VFR (only kept away from other IFR).

xenadu02··on Why LLMs can't really build software
I don't think we fully understand all the aspects of intelligence. What the potential feature set is. How to categorize or break it down into parts. We have some data and some categories but we are so far away from a full description that it only makes sense we must move the goalposts constantly.
xenadu02··on Kodak has no plans to cease, go out of business, or file for bankruptcy
There are plenty of areas Kodak could make a difference if they cared to.

Fuji's instant Polaroid-like camera is a novelty and we still use it. Getting an immediate physical print is entertaining for people of all ages these days.

The security camera business is crowded but mostly with low image quality garbage.

For that matter webcams are also often garbage. A quality camera with optical zoom and good ISP and low light performance could do really well, especially with so much remote work.

The market for high speed cameras is still under-served. As is the market for high quality microscope cameras.

There are lots of opportunities out there if you aren't only focused on products that can sell a billion units.

xenadu02··on Why LLMs can't really build software
Every AI-related invention is hyped as "intelligence" but turns out to be "Necessary but Not Sufficient" for true intelligence.

Neural networks are necessary but not sufficient. LLMs are necessary but not sufficient.

I have no doubt that there are multiple (perhaps thousands? more?) of LLM-like subsystems in our brains. They appear to be a necessary part of creating useful intelligence. My pet theory is that LLMs are used for associative memory purposes. They help generate new ideas and make predictions. They extract information buried in other memory. Clearly there is another system on top that tests, refines, and organizes the output. And probably does many more things we haven't even thought to name yet.

xenadu02··on Someone keeps stealing, flying, fixing and returning this man's 1958 Cessna
Landing the airplane in a way that you don't damage the airplane or yourself is another matter entirely.

You can also easily get yourself killed through a stall/spin, flying into IMC/bad weather, etc.

A lot of pilot training is how to plan for weather, check performance, handle emergencies, and not create chaos for everyone else.

xenadu02··on Someone keeps stealing, flying, fixing and returning this man's 1958 Cessna
Flight plans are only required for IFR flights in many countries. The VFR rules vary (altitude, area, etc). You can file a VFR flight plan if you want to but it is not required.

On the other hand you can't enter Clasa B airspace (the airspace around large airports) without permission from ATC. You also can't fly above 18,000ft in the US under VFR. That keeps small planes mostly away from the big jets.

xenadu02··on US Coast Guard Report on Titan Submersible
To be extra clear here:

They used an unproven custom-designed sensor + controller system to monitor the health of the hull.

The monitoring system detected the beginning of total hull failure in the exact way they intended. They then ignored that monitoring system because hull failure would have been inconvenient.

Really that's the whole story of Stockton here. Massively motivated reasoning. Anything inconvenient was written off as wrong. A lot of normalization of deviance too as some of the written-off concerns turned out to be wrong. But not all of them.

I recently got my PPL and that's a massive risk for pilots. You fly in weather you shouldn't and get away with it for a while. Then it becomes normal. You've "proven" you can handle it. Then a situation comes along that requires more margin than you have left and you die.

xenadu02··on Objects should shut up
On the Kia EV 9 I had to go in and disable lane assist (which you can still disable here in the US).

It was downright dangerous, jerking the steering wheel at seemingly random times when it gets confused.

xenadu02··on So you want to parse a PDF?
PDF is a list of drawing commands (not exactly but a useful simplification). All those draw commands from some JS lib or in SVG? Or in every other platform's API? PDF or Postscript probably did them first. The model of "there is some canvas in which I define coordinate spaces then issue commands to draw $thing at position $(x,y), scaled by $z".

You might think of your post as a <div>. Some kind of paragraph or box of text in which the text is laid out and styles applied. That's how HTML does it.

PDF doesn't necessarily work that way. Different lines, words, or letters can be in entirely different places in the document. Anything that resembles a separator, table, etc can also be anywhere in the document and might be output as a bunch of separate lines disconnected from both each other and the text. A renderer might output two-column text as it runs horizontally across the page so when you "parse" it by machine the text from both columns gets interleaved. Or it might output the columns separately.

You can see a user-visible side-effect of this when PDF text selection is done the straightforward way: sometimes you have no problem selecting text. In other documents selection seems to jump around or select abject nonsense unrelated to cursor position. That's because the underlying objects are not laid out in a display "flow" the way HTML does by default so selection is selecting the next object in the document rather than the next object by visual position.

xenadu02··on Neil Armstrong's customs form for moon rocks (2016)
It wasn't that they didn't have life insurance at all. It was that they couldn't increase their life insurance on the private market either due to their line of work or "Act of God"-style clauses. Or at least they thought that to be the case. This is similar to how some life insurance policies exclude death while acting as a pilot in general aviation.

AFAIK all of them were former military and obviously current government employees so their families would have been entitled to any military life insurance they purchased as well as any pension benefits due (military and federal civilian). I can't give you the exact amounts because it has changed over the years and also depends on how many years of military and/or civilian service you had.

But generally all government employees covered under the retirement plan have an annuity or monthly survivor benefit available that is some portion of their final salary at death or their average salary whichever was higher. Often there is a fixed adder as well (basic death benefit adder right now is $41,000 per year so your spouse would get 50% of your final salary plus $41k).

In addition the federal plan (and social security) pays monthly for surviving children until they reach age 18. The federal plan is a bit nicer in that it pays until 18 or 22 if you are a full time student. Both pay for life if the child is disabled (though the government definition of disabled is rather strict).

All of this is just survivor benefits. Once your surviving spouse retires they are entitled to the pension payments you would have received.

xenadu02··on Unsafe and Unpredictable: My Volvo EX90 Experience
Our leased 2024 Kia EV9 has similar terrible software problems.

Anyone who thinks Tesla's Autopilot/FSD (or any aspect of their software) is bad... much of the competition is far worse.

A few issues:

* Lane keeping gets dangerously close to other cars in turns for no apparent reason * Lane keeping will randomly decide to follow non-existent lanes * You can't turn off lane assist (the baby version of lane keeping) and it tries to override you, leading to jerking of the steering wheel at high speed (eg to avoid an obstacle in the road). * When switching from R to D it wants you to press the brake. But if you are still moving a tiny bit or you don't press the brake hard enough it just shifts you into N instead (!!). I live on a hill and this is only detected when I press the accelerator pedal and nothing happens. But you have to come to a full stop to shift into D (Why???). * Some settings refuse to save to driver profile; to get single pedal driving you must use the paddle shifters each driving session to go from iDrive 3 to Max. But if you are moving too fast it refuses to change the mode. If you set the mode in R it resets when you move to D. * Despite being an EV with key/digital key detection you must manually press the ON button and manually press the Off button. Otherwise when you get out of the car it just sits there ready to be driven away by a thief. * No auto-lock when walking away. * Remember the pedal thing from shifting? Same with pressing ON button. If you don't press the brake pedal down hard enough or give it 1-2 seconds before pressing ON it just turns on accessory mode. * No geofencing so no ability to configure anything to behave differently at home. * Want to control the charge plug locking behavior? Don't bother going to Settings. You won't find it. You must go to the home page, then press the EV Leaf box. Then go to EV Settings from here. There you will find a new settings menu that has the same ones from Settings but it now has a couple of new categories not present with all the other settings of the car. Including whether to lock the charge port door and whether to lock the charge cable into the car itself. * Sometimes in following cruise control mode it just locks in at a speed different from the one you set for no reason. * When you touch the accelerator in cruise it turns cruise off so when you let off the accelerator the car actually jerks you around as it decelerates for a period of time before cruise kicks back in lurching you forward. * Don't press the accelerator for too long or it will just turn cruise control off entirely, including lane keeping. * It wants your hands on the steering wheel but if you move too much it turns off lane keeping but leaves cruise control on. * It has the usual massive plethora of physical buttons randomly scattered throughout the cabin. Some on the center console. Some on the three stalks. Some on the left side where you can't see them. Some below the touch screen. * Different controls behave differently. Sometimes next to each other with similar functions! Opening the rear door? Press and hold. Open the frunk? Double tap the button. The buttons are next to each other. The buttons below the touch screen? Capacitive it seems. Why when the rest are physical? * Despite the cluster being just a huge LCD they do almost nothing with it. The only customizations are for-pay add-ons. * Did I mention the light-up squares on the front are customizable? If you pay for them. Each pattern is an add-on you pay for. * Their app is an absolute disaster. I could do an entire post just about how awful every aspect of it is.

xenadu02··on A 1960s schools experiment that created a new alphabet
We should consider scale in this scenario. What the Chinese did affected millions of people. This story is about a relative handful of kids.

I also find it unfortunate that they bothered with this nonsense. If you're going to do anything just standardize the existing spellings with existing letters. Give long i a standard spelling and be done with it.

xenadu02··on Random selection is necessary to create stable meritocratic institutions
Success in anything (sports, politics, startups) has a luck element to it. By that I mean there were N other people who were just as capable or qualified as you and it was luck that selected you above them.

Luck of being born in the Bay Area and going to high school with people in the startup community.

Luck of not being hit by a bus and spending your critical early career years in physical therapy recovering.

Luck in meeting the right people while volunteering in your local political party event.

Luck in going to a different restaurant and so not getting a cold from a patron at your normal restaurant so you performed your best when a scout was at game three days later.

It doesn't surprise me that a bit of randomness from a qualified pool would pay off.

xenadu02··on Preliminary report into Air India crash released
For the record I don't think the designers of the switch or Boeing are idiots. The switches have guard notches and the throttle quad has metal guard edges to help prevent accidental activation.

As far as we know this is the first accidental dual engine cutoff at low altitude; with just a bit more altitude (not sure of how much exactly) the engine that had restarted and was ramping would have started producing enough thrust to arrest their descent. That makes the margin of "unrecoverable" a lot smaller than you might initially think.

Bottom line is it is worth considering implementing some protection here:

  1. It can be done in software without a lot of complexity
  2. The transition to "air mode" is relatively reliable.
  3. The failure scenario is the system doesn't provide the protection but because the failure we protect against is very rare that is acceptable
  4. It typically fails "safe": allowing shutdown without delay and worst case is a delay in shutdown.
  5. The fire handle overrides delay; if things are going so wrong the delay matters the engine isn't coming back and pulling the fire handle is likely already part of your checklist.
The benefit being elimination of the small window after takeoff where accidental dual engine shutdown is unrecoverable.

Obviously before implementing something like this the proper engineering and failure analysis has to be done.

xenadu02··on Preliminary report into Air India crash released
First I am a pilot. Not commercial or jet rated but I like to think I have a tiny bit more insight than average.

The point of what GI275 does is as a backup instrument you are much more likely to need it when the electrical system fails or is turned off due to fire. Yet if it just remains on until shutdown pilots would frequently forget to turn it off on the ground, resulting in its battery being worn out. Because it is considered critical it delays its own shutdown. Long enough for you to notice in flight but not so long it wears out the battery (which might result in only a few minutes of power in a real emergency).

My entire point was that engine restarts take some time. If both engines eat a blade or catch fire you are screwed anyway so whether or not the fuel cutoff switch does anything at 1500ft is irrelevant. But that is so rare I don't think we have any events on record. So it might be worth inserting a delay - enough to account for standard climb rates to achieve enough altitude to make restart likely or at least possible. The delay would only be for the second engine shutdown and only for time T after going into air mode. And if the system gets it wrong, thinking the other engine is shutdown when it is not pulling the fire handle would override any delay - and pulling the fire handle is part of any engine failure or departed aircraft procedure I know of. In other words you wouldn't even need to change the QRH or emergency checklists in most cases.

I noted that engineering for aviation is complex and everything has failure modes to consider. Privately I went through several iterations of this idea and discarded them for problems with failure modes and complexity. What I proposed is boiled down to the minimal thing that would have saved this flight.

The other thing I'll say is there is a reason the computer will auto-extend some flaps/slats at slow speed even if you put the handle to zero. And there's a reason auto-throttle provides protection. And with the exception of the 737 the computer auto-starts the APU on dual engine failure. And any attempt to deploy thrust reversers in the air is ignored. And stick pushers exist for good reason.

We put in all kinds of measures to override human decisions to prevent mistakes and errors.

xenadu02··on Preliminary report into Air India crash released
There is a relatively short window where dual engine shutdown is unrecoverable. Once you have a bit of altitude (and these jets climb at 2000-3000fpm) you have time for a restart and as thrust comes back sink rate will decrease even on one engine.

My proposal is during this window if dual engine shutdown is commanded don't do it. Treat it like it is happening - show the EICAS message, give the alert, but don't actually do the shutdown until the window has passed. This gives the pilots 10 seconds of startle factor then a bit of time to flip the switch back on.

Single engine shutdown would still behave as today so sure if one engine eats a fan blade shut it down. Not that it matters, the engine computer is going to cut fuel in that case anyway.

Insert a delay only for shutting down the remaining engine and only for X seconds after transition to air mode. A delay that the fire handle overrides.

Just a tiny bit of insurance. There aren't any emergency scenarios at low altitude where engine shutdown works but pulling the fire handle does not. You are coming right back to land at the airport no matter what.

xenadu02··on Preliminary report into Air India crash released
First, the fire handles would override any delay and cut fuel (and other things) immediately.

Second: the window of time where you don't have enough altitude (aka time) to restart is relatively small. So this could easily be a temporary protection.

It is difficult to find exact data on this but restart to significant thrust seems to be in the 30-60s range. If you run the numbers on climb rate and glide time the possible danger zone is relatively small, a few minutes after takeoff at most.

Is this an extremely rare event? Yes. But most other accident causes are also rare, regardless of whether they are pilot error or mechanical.

For example: you might think no pilot would deploy the thrust reversers in flight but system protection errors and/or mechanical failures have conspired to allow it and a bunch of people paid in blood to learn that reverser deployment in flight at altitude was actually unrecoverable - contrary to conventional wisdom at the time. It turned out everyone was flying around with a "kill everyone now" mechanism. In some cases with a much lower margin of safety than previously believed due to the aforementioned "conventional wisdom" that if it happened it wouldn't be a big deal.

Know what else isn't normally a big deal (relatively speaking)? Accidental shutdown of both engines. Because a single engine shutdown is easily recovered and the aircraft can fly on one engine. And dual engine shutdown is easily recovered with a restart if you have enough altitude. But it turns out there's a small window after takeoff where it is fatal.

Somewhat relatedly shutting down the wrong engine in an engine failure scenario is so common they explicitly train crews to slow down and not immediately shut down an engine after failure because rushing just leads to dual engine loss.

xenadu02··on Preliminary report into Air India crash released
It could be defective switch springs, fatigue-induced muscle memory error, or something else. The pilot who did it saying he did not may not have realized what he did. It's pretty common under high workload when you flip the wrong switch or move a control the wrong way to think that you did what you intended to do, not what you actually did.

That said Boeing could take a page out of the Garmin GI275. When power is removed it pops up a "60s to shutdown dialog" that you can cancel. Even if you accidentally press SHUTDOWN it only switches to a 10s countdown with a "CANCEL" button.

They could insert a delay if weight on wheels is off. First engine can shutdown when commanded but second engine goes on 60s delay with EICAS warning countdown. Or just always insert a delay unless the fire handle is pulled.

Still... that has its own set of risks and failure modes to consider.

xenadu02··on Final report on Alaska Airlines Flight 1282 in-flight exit door plug separation
That's true and part of the reason designing for aerospace applications is tricky.

That sort of thing is also one of the legitimate reasons the FAA can have for not adopting an NTSB recommendation. Requiring a seat for small children is one of those calculations. The FAA ran the numbers and assumed some portion of those parents wouldn't fly and of that portion some would drive. Some portion of flights are for physical or emotional health that would not be handled (you can calculate the increase in suicides from things like missing a loved one's dying moments). And of course driving is way way more lethal. So you have to weigh the deaths from not flying plus deaths from driving against deaths avoided if lap children were prohibited.

← PreviousPage 6 of 34Next →