Comcast customer discovers huge mistake in company’s data cap meter(arstechnica.com)
arstechnica.com
Comcast customer discovers huge mistake in company’s data cap meter
http://arstechnica.com/business/2015/12/comcast-admits-data-cap-meter-blunder-charges-wrong-customer-for-overage/
8 comments
All together now: "Comcast Sucks"
Unfortunately the title makes it seem like a systemic problem. That's not clear because in this case it was a mis-typed MAC address. If there's any systemic problem it's that there's no check/verification to close the loop on whose MAC address is whose.
It reflects another issue though which is that MAC address accuracy probably doesn't matter very much in a world without caps (it would make it hard to diagnose issues, but let's assume a relatively good uptime). That means there could be many, many incorrect addresses and Comcast won't find them until people receieve the wrong bill.
I've had an experience where they claimed a cable modem I had purchased from Amazon was their own (from the MAC address on the label). I had to call in several times because I had loaned the modem out to a friend and they refused to activate it because it was an "unreturned rental modem". They're not even able to keep track of their own inventory, I would hardly count on them keeping track of MACs of all modems for the purposes of data caps.
Well, it's a bad process if it relies on that. One would think that this would be automated somehow when the tech comes out to install the modem. Given the error rate of typos and the size of their customer base, it could affect a lot of people.
Don't forget that this data might be relied upon for other things, such as identifying the subscriber account associated with DMCA requests or such.
Don't forget that this data might be relied upon for other things, such as identifying the subscriber account associated with DMCA requests or such.
Agree, there are many reasons to be upset at comcast, but this is was very simple problem to correct, and it's not like they were doing this intentionally. They even gave him three months of overages free, so he wasn't charged anything even if he hadn't figured it out. Are we really reduced to being outraged over clerical errors that are fixed before anything bad happens?
I think it's reasonable to be somewhat outraged due to the fact that there are actually very few end-users who would have had the skills to prove the metering problems. This guy was advanced enough to be gathering traffic statistics at the modem (way beyond 99.9% of users already) and he still got told the measurements were "probably" legit. He didn't get any consideration until, knowing full well that their numbers were off, he unplugged his LAN for an entire week to prove it.
If we're going to give them a pass on accidental billing, you can't make correcting the problem this user-hostile.
If we're going to give them a pass on accidental billing, you can't make correcting the problem this user-hostile.
I'm concerned that there is a cap at all or more specifically there aren't any viable competitors (with a decent network connection without caps).
Comcast has pretty significant overlap with AT&T UVerse and Verizon Fios. Not in all markets, but in enough.
The real problem is that comcast is correct, most people don't use nearly that much data. So even if all the people who are effected leaves, you are only talking about losing 5% of their revenue. It's probably a net-positive in congested areas.
The real problem is that comcast is correct, most people don't use nearly that much data. So even if all the people who are effected leaves, you are only talking about losing 5% of their revenue. It's probably a net-positive in congested areas.
Who said it was simple to correct?
Surely... if they know how inaccurate their measurement system is... They must have used a more accurate system to work that out right? Why not just always use the more accurate system?
Sounds like bollocks.
Anyway I'd hardly call an isolated typo a "huge mistake". This kind of thing happens all the time. Less excusable in this case because why would you need to type in a MAC address? But still...
Sounds like bollocks.
Anyway I'd hardly call an isolated typo a "huge mistake". This kind of thing happens all the time. Less excusable in this case because why would you need to type in a MAC address? But still...
>Why not just always use the more accurate system?
Time and cost. Do you count retransmits, drop packets before the customer sees them? If so there is a whole lot more tracking that must occur, that means more cpu power, memory, and storage. In the case of the ISP, it probably means more licensing cost on whatever tools they use to capture that data.
What is more worrisome is that utilities have standards set and particular rules they must follow in their measurement systems. Do ISPs? I've seen systems that count all packet sizes as 1518 bytes. If you're doing mostly bulk transport that is generally the case. If you're a customer and do lots of DNS lookups and other small calls, your average packet size will be below that.
Time and cost. Do you count retransmits, drop packets before the customer sees them? If so there is a whole lot more tracking that must occur, that means more cpu power, memory, and storage. In the case of the ISP, it probably means more licensing cost on whatever tools they use to capture that data.
What is more worrisome is that utilities have standards set and particular rules they must follow in their measurement systems. Do ISPs? I've seen systems that count all packet sizes as 1518 bytes. If you're doing mostly bulk transport that is generally the case. If you're a customer and do lots of DNS lookups and other small calls, your average packet size will be below that.
>Surely... if they know how inaccurate their measurement system is... They must have used a more accurate system to work that out right? Why not just always use the more accurate system?
That isn't necessarily true. You can make a connection, send a known file and meter the connection on the home computer/router. Then see what the Comcast system measures.
That isn't necessarily true. You can make a connection, send a known file and meter the connection on the home computer/router. Then see what the Comcast system measures.
> Less excusable in this case because why would you need to type in a MAC address?
That is how they do authorization and billing. If there was no way to authorize modems, you could just plug in any cable modem to their system and get free internet. Its also why you have to call them if you switch equipment.
That is how they do authorization and billing. If there was no way to authorize modems, you could just plug in any cable modem to their system and get free internet. Its also why you have to call them if you switch equipment.
They could also use PPPoE for this:
> https://en.wikipedia.org/wiki/Point-to-point_protocol_over_E...
> https://en.wikipedia.org/wiki/Point-to-point_protocol_over_E...
I'd like to see every municipality's weights and measures division take some interest in Comcast's racket. If Comcast thinks bulk packets are no longer too cheap to meter, then perhaps they should realize the true cost of measurement!
Using a non-certified measurement system to conduct trade? That's a paddlin'.
Charging for network management traffic initiated by Comcast? That's a paddlin'.
Charging for MAC headers solely for Comcast's network equipment? That's a paddlin'.
Not providing some sort of usage report so a customer can track down errant usage? That's a paddlin'.
Charging for a packet that subsequently gets dropped within Comcast's network? That's a paddlin'.
Charging for an incoming packet from an address that a user has requested a block on? That's a paddlin'.
Using a non-certified measurement system to conduct trade? That's a paddlin'.
Charging for network management traffic initiated by Comcast? That's a paddlin'.
Charging for MAC headers solely for Comcast's network equipment? That's a paddlin'.
Not providing some sort of usage report so a customer can track down errant usage? That's a paddlin'.
Charging for a packet that subsequently gets dropped within Comcast's network? That's a paddlin'.
Charging for an incoming packet from an address that a user has requested a block on? That's a paddlin'.
> Using a non-registered and non-inspected device to measure for trade? That's a paddlin'.
I wonder if that isn't the best way to approach all this. Not just for Comcast but any carrier that meters stuff out; mobile phone data, mobile phone minutes, etc. If you want to claim I get X amount, prove it!
Now that you mention it, I can't help but wonder why Starbucks doesn't have to "prove" that its 12oz latte really is 12oz. Is it because I have the right to immediately measure it and complain? I don't know much about this aspect of the law.
I wonder if that isn't the best way to approach all this. Not just for Comcast but any carrier that meters stuff out; mobile phone data, mobile phone minutes, etc. If you want to claim I get X amount, prove it!
Now that you mention it, I can't help but wonder why Starbucks doesn't have to "prove" that its 12oz latte really is 12oz. Is it because I have the right to immediately measure it and complain? I don't know much about this aspect of the law.
>Now that you mention it, I can't help but wonder why Starbucks doesn't have to "prove" that its 12oz latte really is 12oz. Is it because I have the right to immediately measure it and complain?
In a different life, I worked at Starbucks. About once a quarter (per my recollection) we got secret shoppers who, among other things, would actually weigh the drinks they ordered to ensure employees were making them correctly. Bad reports were a pretty big deal, so management tended to take care to respond to any problems quickly.
And yes, you could absolutely complain if you ordered a 12 oz latte and received one that was only 3/4 filled. You'd get it remade immediately and probably a coupon for a free drink on your next visit (at least, that was policy at the store I worked at, when I worked there).
Things like scales (for weighing purchased coffee beans, as well as internal use) were actually certified by the county department of weights and measures every year.
In a different life, I worked at Starbucks. About once a quarter (per my recollection) we got secret shoppers who, among other things, would actually weigh the drinks they ordered to ensure employees were making them correctly. Bad reports were a pretty big deal, so management tended to take care to respond to any problems quickly.
And yes, you could absolutely complain if you ordered a 12 oz latte and received one that was only 3/4 filled. You'd get it remade immediately and probably a coupon for a free drink on your next visit (at least, that was policy at the store I worked at, when I worked there).
Things like scales (for weighing purchased coffee beans, as well as internal use) were actually certified by the county department of weights and measures every year.
I'd imagine it's fundamentally because restaurants don't actually use a precision measurement device to dole out drinks. If you saw the cup was low you could complain, and if they used fraudulent cups that said 12oz but were really 11oz, then that would presumably pretty easy to prove (since the now-wise customer would have a cup).
I'd expect state laws to be pretty specific about what types of measurements are regulated, but data usage seems like it would conceptually fit given the opaque and after-the-fact nature of its reporting.
Phone companies have been keeping call records forever, and even printing them line-by-line on paper in the past. The design theory behind IP was kind of to avoid that kind of centralized complexity and its associated overhead, but alas Comcast is hell bent on reimplementing it because decommodification is profitable.
I'd expect state laws to be pretty specific about what types of measurements are regulated, but data usage seems like it would conceptually fit given the opaque and after-the-fact nature of its reporting.
Phone companies have been keeping call records forever, and even printing them line-by-line on paper in the past. The design theory behind IP was kind of to avoid that kind of centralized complexity and its associated overhead, but alas Comcast is hell bent on reimplementing it because decommodification is profitable.
At least in Germany in restaurants glasses have to have a "Füllstrich" for this reason:
> https://de.wikipedia.org/wiki/F%C3%BCllstrich
The "Füllstrich" is a small line at the glass with a number next to it with which you can easily check whether you got the claimed amount.
> https://de.wikipedia.org/wiki/F%C3%BCllstrich
The "Füllstrich" is a small line at the glass with a number next to it with which you can easily check whether you got the claimed amount.
Careful what you wish for. I'd prefer my isp not keep careful records of where my bytes are going and when.
They don't have to keep careful records of where and when. They just have to spend a lot of money buying devices which are manufactured and certified by someone other than themselves that they are fit for purpose.
If the power company isn't allowed to make their own meters then why should Comcast be allowed to? The moral hazard is pretty substantial.
If the power company isn't allowed to make their own meters then why should Comcast be allowed to? The moral hazard is pretty substantial.
All other things being equal I would prefer my ISP to not even have the equipment in place to determine whether I'm downloading from Spotify or Youtube.
I agree completely. But your power meter doesn't know what device in your house is using electricity, it just knows you're using X amount of electricity.
Yes it would be difficult to make a data meter that's intentionally blind to where the packets are going to and from, but it's not as though the data isn't available for intercept a dozen other places on a telco's network. If they want to spy on you and your data is on their network there's little you can do.
The point of this independently certified metering device is that it'd be expensive. Hopefully nobody would sell them, only rent them. So if a carrier wants to charge overages they're more than welcome to, but the people supplying the meters are going to get a huge cut of their business. And once a telco is on the receiving end of a racket instead of on the giving end, perhaps they'd change their tune about data caps and spend that money and energy upgrading their network instead of using it to clamp down on their "customers" who have little to no alternatives.
Yes it would be difficult to make a data meter that's intentionally blind to where the packets are going to and from, but it's not as though the data isn't available for intercept a dozen other places on a telco's network. If they want to spy on you and your data is on their network there's little you can do.
The point of this independently certified metering device is that it'd be expensive. Hopefully nobody would sell them, only rent them. So if a carrier wants to charge overages they're more than welcome to, but the people supplying the meters are going to get a huge cut of their business. And once a telco is on the receiving end of a racket instead of on the giving end, perhaps they'd change their tune about data caps and spend that money and energy upgrading their network instead of using it to clamp down on their "customers" who have little to no alternatives.
I agree, but why would you assume they don't already?
Transporting bulk bits is a commodity race to the bottom, which is why the idea is so technically successful. Now that there's proven demand, telcom raises profit margins by decommodifying opaque bits into specific value-priced services.
Transporting bulk bits is a commodity race to the bottom, which is why the idea is so technically successful. Now that there's proven demand, telcom raises profit margins by decommodifying opaque bits into specific value-priced services.
> Charging for network management traffic initiated by Comcast? That's a paddlin'.
> Charging for MAC headers solely for Comcast's network equipment? That's a paddlin'.
I would love to see them set up a cable modem with nothing connected to it for a month, and then use its data numbers to tare everyone else's meter.
> Charging for MAC headers solely for Comcast's network equipment? That's a paddlin'.
I would love to see them set up a cable modem with nothing connected to it for a month, and then use its data numbers to tare everyone else's meter.
"Comcast has been steadily introducing data caps into parts of its territory, testing customers' responses before a potential nationwide rollout. "
Sadly a lot of people will complain and will dislike this meter, but have no other choice as Oleg in the article and will have to stick with comcast. Comcast will later say we lost 0 customers in areas we switched to metered bandwidth.
Sadly a lot of people will complain and will dislike this meter, but have no other choice as Oleg in the article and will have to stick with comcast. Comcast will later say we lost 0 customers in areas we switched to metered bandwidth.
Well how about that! I logged into my Comcast account and saw we had hit nearly 250-300 GB each of the previous three months.
There was no way (unless my wifi was hacked) that we would use that much. We do have Netflix, but we watch maybe 3-5 hours a month? We rarely stream anything.
So I logged into my router and saw that in the last 30 days we used 23 GB. Much more in line with what I've seen in the past.
Just waiting for Comcast to charge me extra fees!!
There was no way (unless my wifi was hacked) that we would use that much. We do have Netflix, but we watch maybe 3-5 hours a month? We rarely stream anything.
So I logged into my router and saw that in the last 30 days we used 23 GB. Much more in line with what I've seen in the past.
Just waiting for Comcast to charge me extra fees!!
I really hope this ends up resulting in a class action lawsuit.
I don't have Comacast so I can't look it up right now but I bet Comcast contracts include a binding arbitration clause that prevents exactly that.
how unlikely is this, really?
a MAC address is 48 bits or 6 hex digits. what are the odds that mistyping one of those digits resolves to a MAC address of another Comcast customer?
are they assigned randomly?
why are people at Comcast typing in MAC addresses in situations as serious as billing and metering, in the first place?