My father works ATC and his schedule has him working overtime, 6 shifts a week, including overnight shifts, meaning that there is literally not a day of the week where he doesn't spend at least some time in the tower.
If that's the reality for even half of the controllers, it's no surprise that we've been seeing more and more traffic accidents lately.
Overweight/obesity combined: ~73-75% (nearly 3 in 4 adults) in the US.
We get it, @ExtraRoulette. You're big pimpin'.
https://news.ycombinator.com/item?id=46046401
But I don't think we should extrapolate from it.
If so, doesn't the understaffing (lower # of employees) result in each employee being overpaid (paid a higher hourly rate)?
EDIT: And it seems like air traffic controllers can retire after just 20 years and draw a defined benefit pension: https://www.faa.gov/nyc-atc
You can still do contract ATC work after 56.
The mystery to me is that AT shortages have been known fora. while now, so why haven't many more trainees been recruited?
ATC has been a shit career prospect for a while now so no one wants to enter training.
For one it requires uprooting your entire life to live near a training center, then they send you on an apprenticeship to a random airport in the country for a few years. And since there are only so many slots in the desirable metros, most people get sent to live somewhere “undesirable” to say the least.
For two, while trainees get paid they get totally fucked during government shutdowns. Many who make it to the funnel also quit at that point. Without fundamental structural changes to how they’re trained and paid at the political level, the number of trainees will remain small.
You can’t account for overtime in pensions because the employees will conspire to force overtime for retiring employees to bounce the pension up. Just a natural risk with an entity that can’t go bankrupt hiring people.
The money doesn't somehow make it sustainable for the people burning out their lives. Working 7 days a week, including overnight shifts, for 20 years to collect a pension seems like WELL earned compensation.
That's seems unrelated to "we have so few" and "we enmiserate the one's we do have".
It's hard to argue you're underpaid if, as a result of short staffing, you're getting paid more (both in absolute terms and per unit of effort) than you signed up for.
There have been many attempts to change phraseology, teach pilots and controllers to always readback runways, etc. but nothing that actually prevents the issue from occurring entirely via automation.
The pilot was given the clearance to land before the fire engine was dispatched. Apparently there was not even enough time for the crew to max out the thrust and try to lift off the strip even if they managed to notice the lights of the incoming fire engine.
In my job we work 40h a week + oncall rotating. It works.
Why not blame any number of people who held the same office between then and now who have equivalent power to fix the system?
If we assign blame to this dead guy a long time ago, then there is no accountability to be had.
The union pretty loudly and early on pointed out major problems with that job and the response of ignoring them for 4 decades is what's driven us to the current situation.
Multiple economic write ups have concluded that Reagan’s “stick it to the upstart guy” cost us tax payers way more than it would if they’d just acceded and maybe even thrown in a gracious bonus to say thanks.
Larger sociology say the intangible cost to labor balance laws actually were much more.
Reagan’s trickle down (great euphemism for “piss on”) movement was the beginning of the demise of the GOP IMO. Disclaimed: I voted both times for him and many GOP followers.
A union that isn't allowed to legally strike when needed isn't a useful union though. The state that ATC has been in for the decades after that suggests to me that they were correct to strike.
How dare those peons use their economic leverage! That's only for the upper class
Clearly human-run ATC results in situations like this, so the idea that automated ATC could result in a runway collision and should therefore never be implemented is bad.
You obviously wouldn't authorize the bot to do everything, but you could allow it to autonomously call for stops or go-arounds in a situation like this where a matter of a few seconds almost certainly would have made the difference.
Imagine the human controller gives the truck clearance to cross and the bot immediately sees the problem and interrupts with "No, Truck 1 stop, no clearance. JZA 646 pull up and go around." If either message gets through then the collision is avoided, and in case of a false positive, it's a 30 second delay for the truck and a few minutes to circle the plane around and give it a new slot.
If we aren't confident enough in the automation to allow it to make the call for something simple like a runway incursion/conflict (via total automation), augmentation might be worse than the current approach that calls for 100% awareness by the ATC. Self-driving research shows that at level 2 and level 3, people tune out and need time to get back "in the zone" during a failure of automation.
We can probably semi automate runway crossing. Someone mentioned red lights when you definitely cannot cross
You're left with a bunch of planes in the sky that can't stay there forever, and not enough humans on the ground to manually land them.
Now image the outage is also happening at all airports nearby, preventing planes from diverting.
How do you get the planes out of the sky? Not enough humans to do it manually.
Now imagine the system comes back online. Does it know how to handle a crisis scenario where you have dozens of planes overhead, each about to run out of fuel? Hopefully someone thought of that edge case.
One of the failure modes should not be “guy forgot thing”.
Remember when all the Waymos were confused by a power outage? Now do that, but with airplanes that will fall thousands of feet and kill hundreds instead of park in the middle of the street.
I'm not saying we shouldn't automate things. We should. But, it's not easy. If it was, we would have done it already.
It already is.
> Now imagine there's a 3 hour outage of the automated system.
Planes divert to another airport, passengers grumble, end of story. Airport closures can and do happen all the time for all kinds of reasons, including weather or equipment malfunctions.
My suggestion is to restrict the use of smaller jets like crj and turboprops. I know airports like LaGuardia can't handle the big jets either, but they could reduce the slots and require a jet that holds, say, 150 people or more. This would result in fewer flights per day to some airports, but reduce overall congestion while still serving the same number of passengers.
> In audio from the air traffic control tower at LaGuardia, a staff member can be heard saying: "'Truck One, stop, stop, stop!" in the seconds before the crash.
It sounds to me like either the Cop or the Firefighter (whichever was driving) wasn't listening to ATC and this whole incident was probably completely avoidable.
EDIT: a video of the crash seems to have warning lights that the emergency vehicle ignored.
One controller working tower duties, ground movement duties, coordinating with other ATC functions off the radio, an active emergency request, and giving clearance amendments all within 2 minutes. It's insane understaffing. On top of it, there was nobody there to take over after the crash. He worked the whole cleanup for the next 30 minutes.
This is an Olympian level elite Air Traffic Controller who was setup to fail.
I've visited towers, center facilities, and have flying (and some instructing) in the San Francisco airspace for 10 years. That kind of failure is systemic way above an individual.
It seems like less than 2 seconds from declaring intent to cross until they are told not to cross.
The runway entrance lights look red to me which is also a huge warning flag.
Digital comms is available in the US:
* https://www.faa.gov/air_traffic/technology/DataComm
The issue is that the final approach and landing (and taxiing?) environments are probably too dynamic for that: in this particular situation one of the vehicles was responding to an emergency (fire).
In addition to huge planes, there is baggage transportation, passenger buses (to mid-field terminals), fuel pumpers, emergency vehicles, snow plows, deicers, and general maintenance vehicles (clear debris off runways).
Literally the crash here was caused by a fire truck entering the runway.
No one here or anywhere is saying automation would solve or be able to handle everything that human operators handle, that's an argument you invented that no one is making.
People are saying automation could handle a significant portion of the routine things allowing humans to handle the more complex/finicky issues.
Even if automation could handle 10% of the most common situations it would be a huge boon. In reality its probably closer to 50%.
If the reason you have the human there is to handle the unusual cases, you run the real risk that they just aren't paying attention at critical moments when they need to pay attention.
It's pretty similar to the problem with L3 autonomous driving.
Probably the sweet spot is automation which makes clear the current set of instructions on the airport which also red flags when a dangerous scenario is created. I believe that already exists, but it's software that was last written in 1995 or so.
Regardless, before any sort of new automation could be deployed, we need slack for the ATC to be able to adopt a new system. That's the biggest pressing problem. We could create the perfect software for ATC, but if the current air traffic controllers are all working overtime and doing a job designed for 3 people rather than one, they simply won't have the time to explore and understand that new system. It'll get in the way rather than solve a problem. More money is part of the solution here, but we also need a revamped ATC training program which can help to fill the current hole.
Very possibly. It will be interesting what comes from the investigation.
> No one here or anywhere is saying automation would solve or be able to handle everything that human operators handle, that's an argument you invented that no one is making.
I’m asking if it would have solved even the current situation. The truck presumably saw the red light, and was asking to cross. Would traffic control have said no if more had been automated and if so, what automation would fix this? Unless we are supposing the truck would be autonomously driven and refuse to proceed when planes are landing, in which case, maybe, though that’s not really ATC automation anymore.
Software routinely solves database coordination problems with millions of users per second.
Even if most of the work is routine, you definitely still want a human in the loop.
This isn't a Kubernetes cluster where you can add VMs in 30 seconds.
....if they go around kilometer of the runway the fire will turn into bigger fire
I imagine the training will consist of something like changing the comms protocol to say “runway lights are on, control. Truck 1 confirming cross runway 4D?” prior to crossing. Double check so to speak.
A naive view that confuses the map with the territory.
While in a database state you write a row and reality updates atomically....for aircraft they exist in a physical world where your model lives with lag, noise, and lossy sensors, and that world keeps moving whether your software is watching or not. Failed database transactions roll back, a landing clearance issued against stale state does not. The hard problem in ATC is not coordination logic but physical objects with momentum, human agency, and failure modes that do not respect your consistency model.
Or to stick with the language of the analogy, every fruit tree has some fruit that is lower than the others. That doesn't mean all "low-hanging fruit" is within arm's reach of the ground, some fruit just doesn't require as big of a ladder as other fruit.
This comment isn't a judgment of this specific case. I don't know enough about ATC to have any confidence in my opinion on the viability of replacing humans with software.
That was my first comment in this thread, so there was no established goal to change. My sole goal was to clarify the meaning of an idiom that the comment I was replying to was misstating.
I even included a disclaimer that "This comment isn't a judgment of this specific case", so I don't know how you could have received it as such.
Go write it then.
It's stupid, wasteful, and ultimately dangerous to make a human do a machine's job.
It takes all of two minutes of Wikipedia reading for me to understand why this isn’t simple; why it's actually extremely not simple! If you ignore the incumbency, the regulations, the training requirements, the retrofitting, the verification, the international coordination, and the existing unfathomably reliable systems built out of past tragedies, then sure, it’s "simple". But then, if you're ignoring those things, you’re not really solving the problem, are you?
Those are excuses and encumbrances, not reasons. If they are so important, it leads to a question: what existing automated systems can we improve by adding similar constraints?
If these are just "excuses" and not "reasons," then explain how you have determined them as such.
I would like to say, "Because knowledgeable people have explained the difference to me." But again, this has come up before, and no explanations are ever provided. Only vague, reactionary hand-waving, assuring me that humans -- presumably not the same ones who just directed a fire truck and an aircraft onto the same active runway, but humans nevertheless -- are vital for safety in ATC, because for reasons such as and therefore.
There you are doing it in order to avoid engaging with the substance of what people are saying.
There is no substance in the replies. There never is. Only unanchored FUD.
"Because it's always been done that way, and that's what the regulations say," will not be accepted, at least not by me.
(Really, my question is more like why humans will still be needed in the loop in 2036. If we started automating ATC today, that's probably how long it would take to cut over to the new system.)
If these are just "excuses" and not "reasons," then explain how you have determined them as such.
This is just not how complex systems work. N of 1 events happen regularly, which is exactly what makes them challenging.
You simply asserting every scenario has been seen before does not actually make it so.
Confusion is indeed a common side effect of a job done halfway.
Replying: I'm really confused at the point you're trying to make - you declared yourself not an expert in this field, while loudly declaring it's so easy to automate.
Because we've already done harder things. 1000 takeoffs and landings per day equals a trillion machine cycles between events... on the phone in your pocket. It is an extraordinary claim, requiring extraordinary proof, to say that this task isn't suitable for automation.
Why don't you do it then? What am I missing?
I'm not qualified to do it, I didn't say I was, and in any event, I don't work for free. I'm asking for concrete reasons why it's not feasible. Spoiler: there are no reasons, only excuses.
The concrete reason your ideas won’t work is you don’t have any.
It's not my job to explain how to do it, it's your job to explain why it can't or shouldn't be done. The extraordinary claim is yours, not mine.
Remember how we installed traffic lights all over the roads and now car crashes never happen any more at intersections? Truly automation solves all problems.
Hard to respond to an argument of this quality, at least without getting flagged or worse.
I know this was rhetorical but the obvious answer is a complete lack of any actual ideas. “Just automate it” is a common refrain from people who don’t know how to fix the actual issues with any domain.
Remember how we installed traffic lights all over the roads and now car crashes never happen any more at intersections? Truly automation solves all problems.
It sounds like you're not asking anything at all
Just to play it out a bit, are you imagining that a pilot would be reporting a mechanical failure upon descent into busy airspace to some type of like AI voice agent, who will then orchestrate other aircraft out of the way (and not into each other) while also coaching the crippled aircraft out of the sky?
Are you imagining some vast simplification that obviates the need for such capability? Because that doesn't seem simple at all to me.
> I'm asking for concrete reasons why it's not feasible.
The concrete reason your ideas won’t work is you don’t have any.
Why don’t you describe the hypothetical automation you believe would solve the problems then?
My hunch is that either your ideas are already implemented (like GP post who said they need to add red lights at the runway instances, except yeah, they do have that), or they are just bad.
> indignant sputtering and patronizing hand-waving.
Preemptively insulting everyone who might respond to you certainly looks like you’re asking for a real conversation. :|
Your accusation of “patronizing hand-waving” is especially off base considering you literally proposed nothing except “automating”. Hand waving indeed.
Just curious: how many people in this thread know what SAGE was? A $5 Arduino has more computing power than the whole SAGE network. This isn't 1958, so we don't need the 'Semi' part of 'Semi-Automatic Ground Environment' anymore.
It does, the Runway Status Lights System uses radar to identify when the runway is in use and shows a solid bright red bar at every entrance to the runway. I'm curious what the NTSB has to say about it for this incident. From the charts LGA does have RWSLs. I didn't check NOTAM to see if they were out of service though.
In the audio released by the BBC, the fire truck DID get permission from the tower to cross something, I can't tell if it was the runway in question. However, to cross the red runway lights if lit, you normally need that spelled out too something like "truck one, cross four delta, cross red lights". This did not happen on the BBC audio, which could mean one of many things.
The guy was alone operating 2 frequencies, had an emergency of another aircraft going on… is not so easy as many commenters from the armchair are insinuating
(See my other comment below if you're tempted to say something about visibility.)
When cleared across a runway I'm still going to be looking in all directions, and proceed as fast as I can. I also look both ways at railway crossings even if the guards are up and silent.
I also wonder if you're down to a "one controller" scenario if it would be better for there to be once frequency, not a ground/air split.
I'm not saying its easy, I'm actually specifically saying it's such a hard job we should have automated most of it away ages ago. If the only thing stopping an accident like this is an ATC employee, this _will_ happen in the future.
They came up with rail signals long before the idea of a computer even existed. It's hard to believe voice only communication of routes and runway access is the best path forward. Especially when passenger airliners are involved.
Automation is fantastic. We use it extensively in aviation. However, the long tail of 9s in reliable requires constant vigilance and oversight because anything that can go wrong will.
There are so many failure modes with vehicles and planes using the same tarmac that I fail to see how anything would be worth developing here that doesn't eliminate that requirement altogether.
Presumably this is lack of familiarity with this on the part of firefighters.
> If an Air Traffic Control clearance is in conflict with the Runway Entrance Lights, do not cross over the red lights. Contact Air Traffic Control and advise that you are stopped due to red lights. (ex.: "Orlando Ground, Ops 2 is holding short of runway 36 Left at Echo due to red lights").
Airports are highly controlled environments unlike typical motor vehicle roadways and generally the same rules apply for aircraft, vehicles, and equipment on airport surface movement areas. From all sources I can find, if the RWSLs were working they should have been red and nobody should have entered the runway without further clarification from ATC.
In this case, from the available information, the drivers of the fire truck thought they were cleared, and proceeded to cross while a plane was cleared to land. I'm not familiar with ATC ground radio to know if they were actually cleared or not, but it seems clear that that the drivers thought they were cleared.
Investigation reports will give us more details.
CPDLC is already being deployed domestically. It's currently available to all operators in en route segments.
All runway incursions at towered airports are reported, classified according to risk, and investigated.
I wouldn't be so quick to rule out that there's some kind of relatively easy technological double check that could greatly reduce incidents. The fact that we've not gotten there despite years of effort to reduce runway incursions doesn't mean that it's not possible.
But calling a replacement of major ATC functions with software a "simple fix" is a perfect illustration of why this is a bad idea. Nothing about human-rated safety-critical software is simple, and coming at it with the attitude that it is? In my view, as an experienced pilot, flight instructor, spacecraft operator, and software engineer, that thinking is utterly disqualifying.
Besides, there already are a lot of "simple" fixes in place for this problem, e.g. RWSL, which didn't prevent this accident.
> Besides, there already are a lot of "simple" fixes in place for this problem, e.g. RWSL
It'll be interesting to hear why RWSL didn't help, as it is supposedly deployed at LGA.
Yes, I know it probably costs $300k, surely today you can have a $10k ground version.
You could also show every plane on a screen inside the vehicle and have some loud alarms if they are on a collision path.
You could even just display FlightRadar24, still better than nothing.
You would still get permission for the tower, this would not be an allow system, just a deny system.
TCAS on planes is disabled below 1000±100' (~300m) AGL (above ground level).
ADS-B on vehicles is already a thing (and FAA certified):
* https://uavionix.com/airports-and-atm/vtu-20/
There are three categories of runway incursion types: operator/ATC error, pilot error, pedestrian/vehicle. Even if someone 'knows' that they need to "hold short runway 12", they can still have a brain fart and go through the hold short line.
Unless you want to argue that all vehicles taxiing have to operate (SAE Level 4) autonomously?
This is a REALLY hard problem that the US cannot solve alone. It would require extensive global coordination.
Not insurmountable, but this is not something you can easily roll out piecemeal. If even a single aircraft lacks the compatible equipment you're back to the existing system.
Ok, let's not try improving systems, how's that working out?
I feel the same way about close calls on the road, especially ones involving a vehicle and a vulnerable road user like a pedestrian or cyclist. Way too many lives being saved by a person jumping out of the way at the last minute who shouldn't have had to do so, and then cops and bureaucrats shrugging with "well what do you want us to do, the numbers don't show enough fatalities here for it to be worth fixing" and later when someone actually does die it becomes "this is a horrible tragedy that no one could have seen coming, let's focus on thoughts and prayers rather than accountability that could lead to structural change."
Which has always seemed a little nuts, like in case of hit and run it would definitely be possible to take action based on a plate alone, both for police and insurance purposes. Unless the registered owner can point to the not-them person who was driving their car at that time, then it was them. Or it was stolen, but either way there will be a clear paper trail.
All that creates a mountain of work and man hours that any police department in America would likely put on low priority.
Basically our legal system is too forgiving and the possibility that someone stole their car (even if it was a friend) and returned it 20 minutes later exists, and therefore it's on the police to prove it wasn't.
And the law is pretty hard to change since it would change it to 'guilty until proven innocent'.
In the end the air traffic system is a highly complex but also a highly reliable system, especially when you compare accident rates.
I am sure the working conditions of ATC staff might be improved - but being both a pilot and a programmer, I know that there is no easy digitalization magic wand for aviation.
I'm sure the NTSB report will cover why this didn't stop the accident. Presumably either the system wasn't working as-expected, or the fire truck proceeded despite the warning lights since they had clearance from the controller.
The system is only advisory at present, so if the truck did see a warning light and proceeded anyway, they were technically permitted to do so.
1700 incursions a year, and other articles mentioning multiple near misses a week at a single airport [1]. It is safe in practice, likely largely due to the pilots here also being heavily trained and looking for mistakes, but it seems a lot like rolling the dice for a bad day.
>I am sure the working conditions of ATC staff might be improved - but being both a pilot and a programmer, I know that there is no easy digitalization magic wand for aviation.
I didn't say it'd be free. Just hard to believe radio voice communication is the best way to go.
[1] https://www.nytimes.com/interactive/2023/08/21/business/airl...
Traffic lights instead of mad max intersections are better.
Then there's subway Automatic Train Control.
I don't know that Air Traffic Control staff don't have computer systems for establishing which plane owns what airspace. They at least did do it manually already following specific processes, so it can be at least augmented and a computer can check for conflicts automatically (if it isn't already). And, sure, ATC could still use radio, but there could be a digital standard for ensuring everybody has access to all local airspace data. Or maybe that wouldn't help.
Your ground vehicle wanting to cross a runway could have the driver punch "cross runway 5" button (cross-referenced with GPS) and try to grab an immediate 30 second mutex on it. The computer can check that the runway is not allocated in that time (i.e. it could be allocated 2 minutes in the future, and that would be fine).
But, as pointed out elsewhere, obviously some of this is already present: stop lights are supposed to be present at this intersection.
I'm gonna guess that code never went into production. The problem seems easy until you start looking under the hood.
Money isn't the only reason it's so old. The coordination problems are huge. https://www.theregister.com/2024/09/24/us_air_traffic_contro...
There is digital comms with ATC without voice:
* https://en.wikipedia.org/wiki/Controller–pilot_data_link_com...
* https://www.faa.gov/air_traffic/technology/DataComm
But in the highly dynamic environment of final approach, landing, and taxiing, I doubt it would be practical. Unless we want to try autonomous 'driving' on taxiways and runways?
I remember a debate a year or two ago about a plane ignoring instructions (IIRC it had changed frequencies) and had taxiied onto a runway when a plane was landing. Luckily the landing plane saw this and do a go around so nobody was harmed.
In the aftermath, there were similar complaints to yours. "Why can't they just have lights to block planes when a departing or landing plane was using the runway?" without thinking through how any of that works. For a start:
- How do you allocate that a runway is "in use"?
- If ATC does it, what if they fail to turn the system on?
- What if turning it on or off fails?
- What if it gets stuck on or off? How do you fix it? Are there procedures for ATC to override it anyway?
- There are multiple entry points to a runway. What if they're in different states?
- What company si going to sell such a system and accept liability?
- What training requirements will be needed for ATC and the pilots?
- What do you do if a pilot goes ahead and ignores it?
I think people can't think beyond cars. Cars have had unimaginable effort put into them so they can only operate within a certain window. Even then they require maintenance.
But as soon as you scale up to industrial safety and control systems, a power plant, the engine on a ship, etc you will end up with a bunch of controls where the people using them need to be skilled operators and it is essentially impossible to eliminate mistakes with automation and IT systems. You will need overrides. You will need redundancies. You will need to end up doing things nobody has ever considered before and have to rely upon training, education and experience to go beyond the envelope. That's just how it works.
How many runways crossings are there in a year? How much is "1700+" a percentage of that total?
FAA defines it as "Any occurrence at an aerodrome involving the incorrect presence of an aircraft, vehicle or person on the protected area of a surface designated for the landing and take off of aircraft." [0]
Many runway incursions run no risk of any accident, but are still flagged as issues, investigated, and punished if appropriate.
[0] https://www.faa.gov/airports/runway_safety/resources/runway_...
And the cost of investigating 1,700 should be within the budget.
If 1,700 is a minuscule fraction of all runway uses (as it likely is) then investigating it should be a proportionally minuscule amount of the budget.
* https://en.wikipedia.org/wiki/Runway_incursion#Definition
* https://www.faa.gov/airports/runway_safety/resources/runway_...
All incursions (in the US) are tracked:
* https://www.faa.gov/airports/runway_safety/statistics
Given there are ~45,000 flights per days in the US (and so aircraft and vehicles would move hither and fro around an airport for each flight), 1700 feels like a small number.
Human can misspeak or mishear instructions, but if they were communicated and understood correctly (a read back was correct), but the pilot had a 'brain fart' and went forward instead of stopping, how do we eliminate brain farts?
See 5-2-5 for an example:
https://www.faa.gov/air_traffic/publications/atpubs/aim_html...
NOTE- Previous reviews of air traffic events, involving LUAW instructions, revealed that a significant number of pilots read back LUAW instructions correctly and departed without a takeoff clearance. LUAW instructions are not to be confused with a departure clearance; the outcome could be catastrophic, especially during intersecting runway operations.
The older term was "hold short runway X" and that was too close to "hold runway X" - the first meant do NOT enter the runway, the second meant enter and line up but do NOT takeoff.
We talk about Vision Zero for streets. Vision Zero is actually achievable in aviation.
Voice communication is insane? I suspect you are ignorant of what it is like to actually fly a large aircraft into a busy airport. Fault-tolerant and highly available hardware must facilitate low-latency, single-threaded communication with high semantic density in order to achieve multi-dimensional consensus in a safety-critical, heterogeneous, adversarial environment.
There is some interesting research that captures this sentiment and shows how complex a solution might need to be (replace "faulty agent" with "human error"): https://www.sciencedirect.com/science/article/abs/pii/S00051...
I don't doubt that it's a very safe system with enough slack allowing for intentional redundancy. But as it is, some of these controllers seem to be limited by their ability to pronounce instructions, leaving absolutely no margin for error and presumably very little room for conscious thought.
Almost all voice transmissions are routine instructions/clearances from ground to air, with the pilots reading them back to reduce the chance of errors. In fact, this already exists and is in wide use in (at least) the US, EU, and in transoceanic airspace.
Of course, now you have two systems that can fail, and reducing reliance on the older one can easily cause automation complacency (which is a well-researched source of errors) and require more frequent refresher courses if the skill is not practiced on a continuos basis.
I suspect that that these are the reasons it's not commonly used for approach and tower operations: There's a lot more spontaneous and/or nonstandard stuff happening in those flight phases, and as you say you don't want a pilot's eyes on a tiny screen/keyboard instead of on their instruments or out the window.