After two years of working well, my Verizon FiOS 75/75 service flaked out on a Sunday afternoon. Packets were dropped, even before it was possible to resolve host names via DNS. I managed to do an Ookla Speedtest from my phone and got 25/0.05 as the result.
Thus began a two-hour phone odyssey with Verizon tech support that began with a heavily accented person seemingly in a distant foreign land who was plainly reading from a script. He had ideas that were absurd if you thought about it for a few minutes, e.g., rebooting the wired desktop PC after a report that all devices connected either wirelessly or wired were unable to get Internet access. He ideas that were absurd if you thought it for 5 seconds, e.g., unplugging and checking the coax cable (untouched for two years since the install) going into the ActionTec router to see if it was “in good condition” (how would a consumer know?). He was preparing to send a service technician to the house later in the week, warning that if the problem turned out to be my PC rather than Verizon’s gear, we would be charged (amount unspecified). I asked “Do you have anyone there who is familiar with computer networking?”
The networking person (who asked me to tell him what was shown by a tracert) concluded that it was likely a problem with Verizon’s network and filed a trouble ticket. He also shipped out a new router in case the problem did turn out to be the router. By the next morning whoever had looked at the trouble ticket got everything working again.
As we talk about artificial intelligence and the glorious future of self-driving cars where we will be entrusting our lives to software, I wonder why it is a human job to look at network quality. Verizon actually supplied me with the ActionTec router. The router runs, I think, a full Unix operating system. Why doesn’t the router periodically measure network quality and, if there is a problem, use its last few packets of connectivity to alert Verizon tech support automatically? (Or maybe use the landline channel, which continued to work throughout this debacle.)
One of the great things about paying taxes to support the Great Father in Washington’s antitrust bureaucracy is that we get to choose from either 1 or 2 broadband Internet vendors in any given U.S. location. We also get to choose from device operating systems made by one of three companies: Microsoft, Google, and Apple. Why is it my job to sit at a Windows 10 browser and notice that web pages can’t be viewed? Why isn’t it Windows 10’s job to be able to detect a near-total network failure and send out UDP packets to the relevant monopoly broadband supplier? It is a pretty short list of vendors with whom Microsoft would need to agree on a protocol.
[And, separately, as long as we’re talking about AI, why can’t Microsoft Word notice that there is a bunch of non-bold small text interspersed with bold larger text and conclude from this that the large bold items are headings and should be “kept with next” automatically? Or at least prompt the author “Do you want to keep this with the next paragraph?”]
The secret of dealing with any overseas tech support is to get them to elevate to a US support person. Companies like Verizon and Comcast have very competent and knowledgeable people here in the US – they just won’t let you talk to them until you jump thru some hoops first.
I understand why the companies won’t let you talk to their (high paid) US tech staff in the 1st instance. Probably 75% of the calls (if not more) are probably in fact caused by some trivial stupidity on the customer end – a cable that is loose, router unplugged, etc. Those scripts do in fact have some connection to reality and represent the most likely causes of disrupted connections in rank order. Imagine that the person calling is not you but your (non-tech savvy) mother instead. Not only would it cost them a lot of money to have their valuable support staff talking to the ignorant but the support staff would go nuts from having to deal with stupid questions all day. This way the blind lead the blind and when it works everyone is happy. That your time gets wasted being led in circles for a while by some blind guy is not a cost to them.
In your case, there was still some shred of uplink so your router could have phoned home but I think in the more typical case the connection is completely cut so that even if your browser or router was smart enough to know that the link is not working, there would be no way for it to send out the distress signal unless there was a backup communication method.
OTOH, in most places there is cellular reception so to include a sim card in the router and send out a few packets of distress signal on the cell network would cost pennies. Better still, if your cable/fiber link was down, the device would fall back to a cellular connection until the main link was restored. This wouldn’t have to be full bandwith – just a 2g connection so that you could continue to get your email and do basic web browsing would be better than nothing.
9 times out of 10 when I lose connection it is something at the head end and when I call, Comcast already knows about it – they just don’t tell you. It wouldn’t be that hard for them to at least send you some alert (maybe a text message – having network problems – we expect your connection to be restored in X hours) so you don’t waste time on the phone with India doing the idiotic “reboot your computer” dance.
If the customer routers sent keep-alive packets to the operator equipment, then all monitoring could be done by the operator.
I’m looking forward to the day when the home network includes hundreds of devices from dozens of manufacturers (all Chinese). Troubleshooting will be such fun.
My personal favourite is when my older relatives are told to unplug cables and plug them back in again. It invariably turns a 30 minute service provider outage into a week-long ordeal.
I don’t know if phoning home would help much (too easy to fail just when you been it), but maybe a big red cryptic 3-digit error code on the device would do the trick. At least it would be better than glaring blue LED lights next to stylishly invisible labels in a dark corner under the desk: they’re about as meaningful as the lights on a UFO to most people.
it’s really good topic to dissuade about network problems, desktop problems because now a days many new PC’s are getting errors and the best solution is to learn how to fix the errors instead of calling to tech support an wasting our time.
Instead of a red blinkenlight, perhaps the device should show a special one-time VoIP number to call for assistance, bypassing India altogether. Perhaps the same call could attach the various logs and data useful to troubleshoot as well. Leaving the monitoring to the operator is, I believe, technically quite feasible but undesirable because they are not self-motivated to fix problems in a timely manner nor make sure they never appear in the first place.
My experience with these various devices, like routers, is that they are too cryptic and too unstandardized. It’s often impossible to figure out what’s wrong because no logs, no legible errors, no errors at all even. Plugging together these devices can be a pain which frankly seems needless these days. Add to that the terrible security we’re all afflicted with.
Managing a home IoT network (with, say, a couple of hundred devices, each with their own complex standards compliance and homegrown solution that could quietly stop working) seems practically impossible and definitely undesirable, even for someone working in the field, not to mention the average consumer.
Similar problems here in Spain —
After having super-reliable 300Mb Vodafone fiber service at home for almost a year, Movistar finally offered similar coverage at our office, which has proven less reliable, and dealing with customer service has been frustrating.
Whereas in the past, network providers wouldn’t stop as soon as you mentioned you’re on a Mac (“Oh, we don’t support Mac.”) today’s equivalent seems to be, “Oh, you’re using our router’s wifi? Then you’re on your own.”
In fact, Movistar wouldn’t even provide support unless we found a device that could connect to the router’s ethernet port to test from _there_.
People seem to want magic from providers (and maybe someday that magic technology will exist that will enable providers to do remote diagnostics but it doesn’t yet). You have to help them do troubleshooting and troubleshooting is done by process of elimination and isolation. If your router works with a wired ethernet connection but not with a wifi connection, then you know that the problem is in the wifi. Since you are not paying the provider to support your wireless LAN (even if they rent you the router) it’s fair of them to find out if the problem is in the LAN.
If your faucet leaks, do you call the water utility? If a light bulb burns out, do you call the electric utility?
@Jackie — I understand, and agree with your point, but the examples you give are irrelevant, IMHO. If the TV provider promises 200 channels, and I’m only see 20, then I call the TV provider. If the ISP promises 300Mb, and I’m seeing 50Mb, then I call the ISP, even if it’s over their router’s wifi network.
As an aside, but related, our company provides software development services to a company that provides wifi equipment to ISPs who provide wifi-based connectivity. Their wifi devices provide all the telemetry necessary to troubleshoot end-user’s connectivity issues, even into their local wifi-based LANs. (And that’s Phil’s point — that for this kind of stuff, end-users shouldn’t even need be involved in the troubleshooting…)
>Why doesn’t your desktop PC
>or carrier-supplied router complain
>to the broadband carrier when the
>network is broken?
Because that would take a bit of energy and imagination from management.
“If the ISP promises 300Mb, and I’m seeing 50Mb, then I call the ISP, even if it’s over their router’s wifi network.”
Maybe if you are right next to the router that’s (almost) justified but it’s not uncommon for a 300mb wifi connection to degrade to 50mb one or two rooms away . Even if you are right next to it and it’s a ISP supplied router, maybe they should charge you for the call if the problem turns out to be in your device and not their router.
Recently my carrier upgraded my tier from 100 mbps to 200 mbps and when I did the speed test I found I wasn’t getting anything over 100 mbps on a wired ethernet connection. So I ran their troubleshooter:
https://speedexperience.xfinity.com/
At it wisely advised me that I needed to replace my old router with a gigabit router. And I did and now I am getting the full 200 mbps. So they seem to be working on some level of automated diagnostics and I suspect that these will only get better because every support call (even to the low paid 3rd world call center drones) costs them $.
I really appreciate Ramu’s perspective – 1st line phone tech support has a thankless task. Either someone like Phil calls who knows much more than the 1st line tech is paid or trained to know and frustration ensues or else (much more commonly) someone who knows nothing calls and the problem is something very basic (and often having nothing to do with the supported device or service) and frustration also ensues. And the tech is not permitted by his employer to escalate until he has been thru his entire script, which only increases frustration.
Automated diagnostics may offer a partial way out of this but unless the diagnostics include a robot who can do things like plugging and unplugging cables you are still going to need the end user’s participation to some extent.
I can’t contribute to your technical discussion, but I had a huge LOL moment:
Tech: Do you know anything about computers?
PhilG: Yes, a PhD in EE…
Phil,
I always assumed tech savvy people all did the same thing with the whole check the cables drill. Say “wait a minute”, put the phone down. Make a cup of coffee, return to the phone. Then “okay, what now”. Repeat ad naseum. 🙂
Bart.
I think the right way to think about artificial intelligence is similar to how we think about naturally occurring intelligence, which is that we don’t expect much from naturally occurring intelligence, and when the lowest of our expectations materialize, we call it naturally occurring stupidity.
The tech support person was human and he didn’t exactly shine in that phone call. Why should an artificial device running Unix or Windows or whatever be any different?