Comcast customer discovers huge mistake in company’s data cap meter
arstechnica.com
arstechnica.com
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'.
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'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.
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.
> 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.
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.
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.
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.
> 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.
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.
If we're going to give them a pass on accidental billing, you can't make correcting the problem this user-hostile.
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.
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!!
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...
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.
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.
> https://en.wikipedia.org/wiki/Point-to-point_protocol_over_E...
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.
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.
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?