I don't quite understand the error this article is pointing out and how it could lead to a four fold inaccuracy. Can someone try explain with an example?
They used a 0.25 constant in a formula and the article is arguing they should have used 0.945 instead (difference of ~4x). The constant the Trump administration used seems to have been the one for estimated effect of tariffs on retail prices (ie, consumer facing) and it was more appropriate to use the one for the effect on prices at the border.
I assume that is similar to me being a computer company importing chips, a 25% tariff on microprocessors will make the imported cost of microprocessors go up by ~24% (because some really expensive ones don't get bought) but it would increase the price of the computers I build by ~5% because most of the computer isn't a microprocessor.
It is an exciting time to learn about the country of Lesotho.
My understanding is they are saying the elasticity of the price (what can be absorbed by tariff increase) comes from the transaction in the retail market it ultimately sells to, yet the "equation" was using elasticity of the import costs.
In other words, Trump's equation thinks that import price will absorb (elasticity) the tariff's impact, but that's not true. It's passed on almost entirely (the 0.945 number, so 0.055 may be absorbed).
So it's using wrong numbers to justify the effective tariffs. IMO, I don't think Trump cares. I think the equation was a "straw-man".
Keeping android open source would be a helpful antitrust defence when some of Alphabet / Googles other products dominate markets (chrome, search). I wonder if they're now less worried about anti trust under the current administration.
Google has had a strategy of keeping some parts of Android open source for the benefit of the world, and other parts closed source inside Google Play Services. However European regulators continue to rule that Google Play Services itself is the leverage Google uses to lessen competition. If regulators have antitrust concerns with just one proprietary blob, there's really no incentive to make Android truly open. Opening up the rest of the Android literally doesn't help with antitrust.
These would have also been terms that HP agreed to or even proposed. It's also a taint on their brand that they're not willing to take care of Humane's customers with even a refund for recent customers in a 116mil transaction..
This is wild to listen to.
A) this is a busy atc channel and it's amazing how much complexity is coordinated over noisy radio.
B) within minutes of the accident happening (at 11:48) the ATC controller is calmly asking helicopters in the air if they can assist in search rescue operations asking, if they have search lights and direction them. This is whilst diverting and grounding flights.
I've been told the noise on the recordings is always a lot worse than in practice due to the location of the recording antenna not being ideal. I have no idea if this is correct or not. Maybe someone can enlighten this.
Yep. Radios are much clearer in the air. VHF is mostly line of sight and planes have the benefit of altitude. Approach/center comms can get a bit scratchy at times mostly due to distance, but approach/tower is almost always perfectly clear.
Yes, having tried to listen to ATC with an SDR— the public recordings are made from antenna that placed near the airport, while the official recordings are made from the feeds that the ATC is listening to— and those antenna have their lobes pointed toward the sky.
I watch a lot of pilots on Youtube. Most of them use some kind of audio splitter to record sound. You hear what the pilot hears. But minus the plane noise, vibrations, etc (unless that's mixed in via some open microphone). So, it's actually better.
So, it really is that bad. Especially when flying low, on the edge of the range of the radios, etc. VHF radio is ancient technology. It's not great even under ideal conditions.
If your phone calls sounded that bad, you'd want to upgrade your phone or switch operator. Really not acceptable for consumer grade communication. Not even close. This is how phones sounded half a century ago. "can you hear me .... are you there ..... can you talk louder ....". My grand mother never really unlearned 1930s phone etiquette. Phone calls with her were short (cause expensive) and she'd be shouting at you because that's what you did. Be brief, loud, hang up as soon as possible. That's what pilots still do. My grand mother was born during WW I (not II) and she's been dead for nearly 20 years.
The aviation world is very conservative with new/better/any technology. They stick with "what works". Even if arguably it barely works. Like pretty much the vast majority of radio exchanges between pilots and controllers over VHF radio. Learning how to talk on the radio is the second hardest part of getting an instrument rating. The only thing that's harder is flying a plane with zero visibility ... while suffering extreme information overload because of the constant radio chatter and dealing with confused/stressed/pissed off controllers who have to juggle you and 20 other planes.
A lot of radio exchanges are routine exchanges that involve very basic information: call signs, transponder codes, codes for approaches/departures, courses, altimeter settings, altitudes, weather information, etc. And then a lot of double checking the other side heard correctly, repeating back what you heard, etc. Radio communication compensates for the lack of a more robust/sane way of exchanging information. There literally is no way for a controller to send you information in written format. Other than using their private phones. So, it all gets spelled out over low quality VHF radio.
There's no good technical reason to not do something vastly more reliable, less intrusive, and less error prone. Some might say safer. Many planes have star link connections these days. The passengers get better connectivity than the pilot. They could be having video calls with the controller with crystal clear audio from anywhere on this planet instead of yelling to them over VHF only when they are in range. They could be sending each other emails, documents, digital maps, and lots of other machine readable information, etc. Not a thing.
It's more a challenge of agreeing on what such a vastly better thing would be and then updating each and every plane and tower, airport, controller world wide with equipment that supports that. Add design by committee to the mix and the "state of the art" is something that would make any competent engineer born after 1970 blush.
YouTube is not a good representation of what you hear in a headset. Half the time the mic taps aren't impedence matched and sound like shit.
Besides that, most "controlling" (i.e. instructions that require action) comms have a handshake of sorts - the pilot repeats back the instruction for the controller to confirm. They don't just bark out instructions and hope you heard it, they listen for you to read them back, and will keep doing so until successful. So, even in cases when the comms are degraded, you will go back and forth until there has been positive confirmation (readback) you heard the instruction correctly (or they'll just give you a new frequency).
> A lot of radio exchanges are routine exchanges that involve very basic information: call signs, transponder codes, codes for approaches/departures, courses, altimeter settings, altitudes, weather information, etc.
They are also in a fairly standardized format with reserved "keywords" for many important aspects of operation ("hold short", "line up and wait", "cleared to X"):
Honestly a lot of those recordings probably do sound a lot worse than the radio audio. I rarely see YouTube recordings that are as audible as the radio is in person. Here's what seems to be the reason:
The common splitters you can buy to break out your headset connector to a recorder don't account for the signal level, which is very high on the headset connectors as aviation headsets are traditionally pretty high impedance. I've struggled with the headset audio blowing out even a fairly nice recorder (a Zoom) with the gain turned all the way down. Phones and pocket reporters don't even offer a gain adjustment and the recording can end up almost unusable. In a lot of YouTube videos of pilots, even most, it's pretty obvious that it's clipping at the recorder the whole time. Perhaps even worse, a lot of people are using like a GoPro with auto-gain, so you get the worst of both worlds: it ramps up gain until the noise is loud, and then when someone starts talking, it ramps down the gain, but a little too slowly, and it's still clipping even when it hits minimum.
It's not very common to have trouble understanding the controller when in the air. AM radio at line of sight (as is the case when you're flying) is pretty robust. Maybe the hardest thing to understand are helicopter pilots because there's often rotor noise on their end.
>>>They could be having video calls with the controller with crystal clear audio from anywhere on this planet instead of yelling to them over VHF only when they are in range.
Your forgetting about latency, which matters when you are trying to land an airplane. Also who cares if you could talk to the aircraft from anywhere in the world with different technology? The only thing that matters is there is a low latency audio connection between the traffic controller and all the planes coming in. Direct line of site VHF communications is better in every conceivable metric then any alternative. It has nothing to do with old technology or design by committee.
Besides the "we already have it everywhere" bit, the big advantage of analog AM over UHF/VHF is that it degrades fairly gracefully. As you're getting too far away the signal gets harder to pick out from the noise, but it's not an all-or-nothing digital signal.
That doesn't matter much over Washington DC, but when you get out towards the western half of the country the transmitters are a lot more widely spread out. There's mountains in the way. There's limitations to how low you can be and still be reachable, which sometimes has to be balanced against how high a GA plane can comfortably fly, or oxygen requirements for the occupants.
A better sounding "modern" system is generally going to be worse at handling those marginal situations, which would probably require building a lot more radio outposts in fairly remote areas to compensate.
But the big problem with requiring anything new is getting it into the existing fleet of thousands of decades old certified aircraft. You need a new radio stack. You probably need new antennas. Changing anything on certified aircraft needs tons of paperwork and things like Supplemental Type Certificates for each individual model of aircraft that make it cost 5-50x what you'd think it should cost. Military aircraft are probably 10x worse beyond that.
A handheld COM radio is maybe $200 from Sporty's. Take basically the same thing but package it as a basic Garmin COM radio (GTR 205) and it's now $2,300. If you want a NAV radio in it too (GNC 215) now it's $5,400. Add GPS and ADSB-Out (GNX 375) and now you're at $9,000. You can buy an entire currently airworthy (really old) plane for maybe $30,000.
For some uses you don't even have to have any radio or transponder/ADSB installed on your aircraft. Some aircraft don't even have an electrical power system to run one. Granted they're not allowed in the middle of Washington DC, but still trying to require the entire fleet gets fancy new digital radios would be a monumental challenge and fantastically expensive.
There are some existing better ways to communicate. Larger planes usually have CPDLC (Controller Pilot Data Link Communications), which is basically text messaging from ATC to the plane avionics. That's how they receive their IFR clearance at major airports. In an airliner you're not generally reading it out over the radio and scribbling out your "CRAFT" acronym on a scratchpad like you see on YouTube. (Even your 4-seat steam gauge Cessna can do this via PDC (Pre-Departure Clearance) at supported airports with something like ForeFlight on an iPad).
CPDLC can also be used in the air to communicate less time-sensitive things like altitude changes and reroutes, talking to the operations department of the airline, etc. There's no reason you couldn't put this in smaller GA planes and towers, other than cost.
Airliners that do transatlantic and -pacific routes also have satellite communications instead of using old-school HF radios (which are even worse than VHF).
Given DeepSeek's open philosophy I wonder what their response is to simply being asked for access to the code and data that this project intends to recreate?
While I'm also interested in this, I guess there is value in independent replication as well. Assuming this is doable - and I wouldn't know.
Does anyone know how difficult it is to perform this kind of reproduction? E.g. how much time would it take (weeks? years?) and how likely it is to succeed?
Interesting, so they wouldn't want to disclose something that shows they've illegally (terms / copyright violations) scraped research databases for example.
Won't this eventually come up in legal discovery when someone sues one of these firms for copyright infringement? They'd have to share their data in the discovery process to show that they haven't infringed..
Some people believe they can dodge copyright issues so long as they have enough indirection in their training pipeline.
You take a terabyte of pirated college physics textbooks and train a model that can pose and answer physics 101 problems.
Then a separate, "independent" team uses that model to generate a terabyte of new, synthetic physics 101 problems and solutions, and releases this dataset as "public domain".
Then a third "independent" team uses that synthetic dataset to train a model.
The theory is this forms a sort of legal sieve. Pass the knowledge through a grid with a million fact-sized holes and with enough shaking, the knowledge falls through but the copyright doesn't.
I'm suprised they can't ship (flat) packaging that could be used in Arizona with a simple assembly line.
If they had that packaging design then for this to make financial sense the two way shipping (and loading, unloading, custom clearance etc) would have to be less than shipping the packaging, the setup cost per unit cost of putting the chip in a box
Wait, wait. In the context of semiconductor manufacturing packaging does not mean what you think it means. It is not putting the product in a paper box.
It is about cutting the wafer into individual chips, wire bonding the silicone to pins, and covering the whole thing with epoxy.
You are correct. I was just illustrating what kind of processes belong to the umbrella term "packaging" in the context of semiconductor manufacturing. Was not talking about what particular process are missing from the Arizona facility.
But you are right on that it is CoWoS which is the missing ingredient.
There are 739 comments in this thread and only one of them mentions emotion. That comment is suggesting the appropiate emotion is gratitude.
I'd go further and suggest you take your time and resources to learn about your emotional state and regulate your emotions so you can help others do the same. Find what you want to be a student of from that calm, content emotional state and with less attachment on how to live your life you will likely find more clarity.
reply