- Modern software is way too overcomplicated to take seriously.
- 911 is handled too specially, leading to oversights by implementers.
- Feature importance to you has nothing to do with implementation difficulties.
The only reliable way to test 911 features is a test lab, to which the average engineer doesn't have access to. On top of that, calling 911 isn't exactly as placing a normal call - so the only way to test is... to call 911.
Again, a test lab should help towards these things, but I doubt Google has one accessible to the average engineer working on the dialer. Plus, they most likely don't have a way to automatically test these changes - or they might happen as part of other "features" (remember the Microsoft Teams bug that caused similar issues?).
In the end, the smarter our smartphones become - the dumber they are at doing the one single thing they were initially meant to do - get help in case of an emergency.
And why not? It seems like the real real problem is #4: Management doesn't take seriously people's need to reach emergency services because it's not a profit center.
Considering the amount of teams working on the Google Dialer, and the fact that they might be distributed across multiple cities / countries - this sounds very expensive.
They noticed cell phone reviews over time have dropped signal or quality testing. Which makes sense since each reviewer is just testing in their local environment with no standards or baseline so they were pretty meaningless anyways.
Are they going to be able to test enough phones and wifi/radio equipment to make it worth while? Not too sure.
We've been talking about the cost of making a screened room. "Turnkey" ones cost about $20-50k, but you're going to pay contractors a fair bit beyond that. Not to mention whatever equipment you're putting inside.
https://www.ramayes.com/Refurbished_Radio_Frequency_Shielded...
c.f. my comment:
> > > It shouldn't be too expensive, but
People just don't read and understand that you can agree with the broader scope of comment but not the details.
Yes, so being able to use the phone UI is important for qualitative testing.
> I don't understand why you're trying to shift the blame away from Google. That's just stupid.
I feel like you didn't read my comment. I stated that a screened room isn't prohibitively expensive, but can't be delivered for the suggested $10k. In another comment, I said "Society spends a few billion dollars a year on E911 and related infrastructure. To have faith in it eroded because of a little bit of slipshod validation by phone vendors is a false economy."
Why isn't it possible to have nuanced discussions-- to condemn Google's failure here, but also to suggest that the capital expense to give people realistic test environments is probably higher than others are suggesting?
Frankly, you're being abrasive.
No, you're trying to find a way to make it sound impossible. But what you're saying is more along, probably, "finding a parking lot to test brakes is impossible even for a car company". Something real stupid as that.
But you're too boiling over in vitriol to understand nuance. Somehow, in your mind, saying that you can't do it for $10k appears to be me defending Google.
And if you don't like that option - you can also place test calls to 911 [2] by calling their non-emergency number and arranging a time. It would be easy to perform a once-a-month test call.
[1] https://jretest.com/product/jre-1812f-forensics-analysis-enc... [2] https://www.911.gov/calling-911/frequently-asked-questions/
A screened room with a few devices and a simulator seems like a good resource to have, and it's a reasonable thing to procure at a center or two if they don't have it. It just costs several tens of thousands, not $10k.
It seems like at this point 911 usually works on Pixel, and there's probably even good unit test and automated integration test of a lot of the components-- prearranged test calls with a given carrier's 911 impl isn't likely to fix it. But "usually" isn't good enough: Google needs to actually figure out what's going on, no matter what.
To have faith in it eroded because of a little bit of slipshod validation by phone vendors is a false economy.
Alternatively, I think you can somehow connect the device antenna to your equipment, so that the signal doesn't even have to be transmitted over the air (or at the scale of a femtocell). Still, rather "expensive" to setup in multiple locations for just a team that develops a dialer.
Additionally, you'd have to test this across multiple OS versions, and devices. Still doable, but most likely not incentivized by the managers at Google
And selling phones that can't pass certification is just irresponsible. If you can't make a product work, you're free to be responsible and cancel a product. It's on Google to do businesses legally.
I suspect that the problem must be more complex than a bug in just the phone or the OS; for one thing, the problem is intermittent and doesn’t affect every phone. Still, the lack of action from the FCC is disturbing.
You should be able to use 933 to test emergency services. https://support.bandwidth.com/hc/en-us/articles/210291778-Th...
And by that I mean the emergency mode of the modems
Ultimately with a system like 911, you will always have to do final testing on the real deal. Because this is just too serious to get wrong.
The ironic thing is personally I only upgrade my phone every 5+ years and would be totally happy with longer development cycles.
but if that would be my implementation I would track call in some very simple way and if that's second or third try to call 911 within hour then handle it as ordinary call without that extra functionality as a safeguard
For what it's worth, every telco switch upgrade I performed ages ago, early days of GSM the first number I tested was 911. I made sure the dispatcher could hear me. I don't know whats going on with the phone development side of things. That seems like a QA and customer feedback review problem. It probably also does not help that wireless vendors are slow/hesitant to update phones. There is a fear of bricking phones and customer support nightmares their words, not mine. I could flash update a phone over the air but this was in the 90's. No idea what that process looks like now.
There's enough 911 calls for everyone, they're just not distributed equally.
Software engineers will fight tooth and nails to keep their privilege is being called engineers whilst having none of the responsibility when it comes to the harm they're causing.
Stop trotting out the same baseless comment over and over.
Let me know if find answers to my questions.
And many software engineering contracts contain liability clauses.
I don’t think things will change until corporate management changes (via being forced to, or otherwise).
I don't understand why these devices can't even do the most core feature of a phone properly.
This will sound crass but the development teams (right up to CEO's) should be dragged out to the gallows and flogged.
If it was my product I would have made damn sure the 911 experience was perfect before shipping, and not rested for a minute until any bugs were solidly quashed - up to and including recalling all sold units and overhauling the flakey architecture if needed.
This is life safety we're talking about, not only for their users but also everyone else impacted by their blatent abuse of the emergency services system. Would we tolerate bridges that collapsed with equally ambivalent consequences for those who engineered them?