Malor’s Techdirt Profile


About Malor

Malor’s Comments comment rss

  • Jan 8th, 2016 @ 7:10pm

    (untitled comment)

    In other words: with electrical and water service, the delivery method is pretty much irrelevant. You buy some number of watts or some number of gallons, and you pay for those watts or those gallons, and it's up to the provider to build enough infrastructure to deliver what you want to buy.

    With an ISP, you buying the infrastructure, not the product. You're buying the pipe, not the bytes. The product comes from other entities entirely.

  • Jan 8th, 2016 @ 7:07pm

    (untitled comment)

    Your electrical wires to your house can usually support about 200amps (US / Canadian standards). Having that capacity does not mean that you can run endlessly at exactly 200amps.

    With electrical service, you pay per watt because each watt has a hard cost.

    With a network, you're buying bandwidth, not bytes.

    You're buying an amount per second, not some fixed amount of data.

    Whether you use your amount per second or not, the cost is the same to any reputable Internet provider, because they will have the backhaul provisioned for your use. Like you, they pay for the size of the pipe, not the bytes transferred.

  • Jan 8th, 2016 @ 4:39pm


    >isn't essentially never

    Argh, I wish we could edit. Is essentially never. Sigh.

  • Jan 8th, 2016 @ 4:37pm

    (untitled comment)

    You are making the common mistake of assuming that the speed of your connection entitles you to every possible byte you could jam through there. Reality says otherwise.

    We're not making a mistake. Providers are lying.

    If Comcast sells you a 40 megabit connection, you have every right to expect 40 megabits per second, 24/7/365. If they can't provide that to you, then they lied.

    Here's what they're *really* selling: maybe 1 megabit guaranteed, burstable to 40. This is a huge difference, but that's not how they sell it.

    If we keep them from lying in this way, most of the other problems disappear. Force them to sell a commit rate, and then force them to live up to the commit. Burst rates are useful, but they're not a measure of network quality, and people would figure that out very quickly in a market where ISPs were forced to sell commit rates.

    You absolutely have the right to expect to get what you bought. Comcast absolutely has the responsibility to provide it.

    Truth in advertising isn't exactly a contentious issue. That basic requirement is imposed on every other industry in the United States, and presumably in Europe. Why wouldn't it apply to sellers of bandwidth?

    Bandwidth is like water.

    No, bandwidth is not like water. I'm really starting to wonder where your paychecks come from, because you sound an awful like someone with a paid opinion.

    Water is a limited resource. There's only so much of it. The limit on water isn't essentially never the size of your pipes, it's the amount of water that gets put into the system to begin with.

    Bandwidth is exactly backward to that. Bandwidth itself is free (or so close that it's hard to distinguish from free); the only limit is the size of the pipe. The limitation on bandwidth is immediate demand. It's not like water or power, which have to be created and pushed into the network, at great cost.

    When you use power and water, you're costing your provider significant money. Their margins are pretty low, typically about 10% most places. (If you use ten cents' worth of power, you probably cost your power company about nine cents to deliver it to you.)

    When you use bandwidth, you cost your provider zero dollars. The only impact your usage has is on immediate demand, which may force the provider to build new pipes to satisfy that demand. The only time your usage matters is during peak hours; during offpeak, it has literally zero impact on their bottom line. With utilities, when you get charged ten cents, you cost them nine cents. But with Comcast, they want to charge you ten cents for something that cost them zero cents.

    So, what's the solution? Truth in advertising. Just the truth. It's amazing how useful truth can be.

    If your neighborhood has a 10 gigabit pipe, and it's split 2000 ways, then the provider has about 5 megabits per customer. The commit rates that get sold had better not exceed 5 megabits each by all that much. Most providers, even big ones, oversell, but real ISPs know that those sales commitments were promises that they need to keep, and if they blew it by overselling, then they have to suck it up, and build the infrastructure to make it right.

    They're the freaking experts, not the end users. It's their responsibility to get it correct in the first place, and fix it if it's wrong.

    That would remove any sharing / overselling and would require a major upgrade from the ISPs to handle this amount of traffic properly and correctly.

    So, in other words, ISPs would have to stop lying about what they're selling?

    I could live with that.

  • Jan 7th, 2016 @ 11:33pm

    (untitled comment)

    So do the math, you have to spread the cost of each connection over a certain number of customers. Like it or not, there is a limit to connectivity.

    Sure, absolutely. But when they promise their customers 40 megs a second, don't you think they have an inherent responsibility to actually provide 40 megs a second?

    You've internalized overprovisioning as okay, instead of being something that's only allowable up until it impacts customer commitments. You think it's okay to lie to customers about what they're buying, and then not actually provide that service when they ask for it.

    They paid for 40 megabits. That's what it says on the tin. They have the right to expect that much bandwidth. Comcast has a responsbility to deliver that 40 megabits whenever the customer wants it, because that's what they said was on offer.

    If they can't do that, they are lying about what they are selling.
    Now, in the same manner that you cannot build highways to accept all of the rushhour traffic not matter what, it's not realistic to assume that an ISP will buy bandwidth and connectivity to a level that supports the potential 100% use of all of their clients at any peak moment.

    Okay, that right there, is the center of your argument, and it is bullshit. It's not ordinary level bullshit either, it is epic bullshit.

    Comcast sells bandwidth. That is what they do. If they can't provide what they are selling, then they are lying.

    And you are arguing, here, that it is okay for them to lie. Further, that it's just *expected* that they lie. Marketing trumps reality; customers won't pay for accurately-labeled packages, so it's okay to lie to them so they'll buy something they can't actually get.

    The sheer absolute *gall* of that argument is just... fuck me, it is wrong wrong wrong wrong. It is ethically bankrupt and technically wrong.

    Comcast absolutely can meet its commitments. It absolutely, 100% can. All it has to do is stop lying. It's really that simple. It can either be honest about what it's actually providing, or it can just suck up the cost of providing what they've already promised.

    What's happening now is that they lie about what's for sale, and then punish customers that actually use what they think they bought.

    In the end, a cap isn't really a speed limit, as much as it's something that explains a difference between the size of the "series of tubes" and how long you can hold that tube open for each month.

    Again, absolute bullshit. The only thing that matters, the ONLY thing, is peak demand. Bandwidth caps are not related to peak demand. Offpeak usage costs literally zero dollars, but you're saying that it's okay to charge for something with zero cost.

    Here's how to actually fix the problem for real: make Comcast be honest. Make them sell bandwidth the way it actually works. "Okay, John Smith, we promise to give you 5 megabits anytime, and up to 40 megabits if network conditions allow." In other words: commit and burst, the way that all networks fundamentally work.

    This would actually fix the problem. For real. Permanently.

    Your solution is about piling lies on top of more lies.

  • Jan 7th, 2016 @ 8:25pm

    Re: Re:

    This is absolutely false. It seems like it would be true on the surface, but reality kicks in pretty quick and you start to realize it's not just true.

    There are two problems: All pipes physically have two ends, and it costs to be in both of those locations. Even if the telco or ISP owns the spaces, there are costs related to being in that space.

    Sure, okay, there's a fixed overhead to having a link up. But that overhead does not change with usage. If a circuit costs $X to maintain, it doesn't cost 0.5X when it's lightly loaded and 2X when it's saturated. It's just X. Usage doesn't impose any additional costs. (there's a *tiny* power delta, but it's so small it would be quite difficult to even measure.)

    Whatever your overhead is with a pipe, it doesn't increase based on usage. You build your circuit, buy your router(s), and potentially buy a commit from a transit Internet provider, and then your overhead stays the same whether it's at 100% or 0%.

    You can buy flex transit bandwidth, where it actually does cost you, but, again... the only time you will actually pay for it is *during peak hours*.

    Only peak demand matters. Offpeak usage does not, and bandwidth caps are completely unrelated.

    It's like trying to reduce rush-hour traffic by imposing mileage limits on cars. If people are only allowed to drive 300 miles a month, when are they going to drive? During rush hour, the only time, barring accidents, when traffic matters.

  • Jan 7th, 2016 @ 5:06pm

    (untitled comment)

    You got closer than most at explaining what a pile of crap bandwidth caps are, but I think you still missed the target by a little bit.

    The thing about network connections is this: they cost a lot to build, and have some fixed overhead per month, but usage of a connection costs basically nothing. Once a pipe is built, the monthly cost is the same whether it's fully used or not used at all. (There might be a tiny power delta for heavy use, but it's so small that it disappears in the noise.)

    So all Internet pricing, at the core, works around that fact, that building pipes (and paying for the loans to build them) costs a lot, but usage is unimportant once it's built. If you're a big central ISP, you lay in big pipes, probably taking on loan payments, and then sell pieces of those pipes to downstream customers. You chop it up and resell it, and you don't care how they use it. Fully used, not used, it doesn't matter, if you're properly provisioned to deliver what you've committed to.

    What Comcast is doing, by imposing bandwidth caps, is charging for a product that costs them nothing. Usage never directly costs them anything. This is important to internalize. If their network is at 100% saturation, their costs are the same as if it's at 0% saturation. So charging per byte is absolute, utter mendacity.

    This is what does matter: peak usage. For three or four hours a day, usually in the evenings for a residential ISP, the network approaches maximum load. It's that usage, and only that usage, that can cost them money. Why? Because if they get oversaturated, and start dropping packets, customers will be unhappy. So they have to add more pipes, and the associated monthly overhead, to deal with peak load. That can cost a lot of money.

    But once they've built a connection, and taken on the monthly bill to run it, usage of that connection doesn't change what they pay.

    What it all boils down to is this: Comcast doesn't want to spend the money required to provision properly for peak usage. They're not delivering what they promised, and they're trying to monetize offpeak usage, which costs them zero dollars, by imposing bandwidth caps. The only reason they can get away with this is because people can't switch to other providers. They have physical monopolies, thousands of small ones in communities all over the country, and they're using those monopolies to extract rent.

    At core, that's what is fundamentally going on here. This is a monopoly, and it's abusing its privileged position. It is charging for a service that costs nothing to provide, simply because it can.

  • Dec 29th, 2015 @ 3:57pm

    Re: The Power of Taxes

    $95/mo for gigabit Internet isn't quite as cheap as Google Fiber or EPB, but if they follow the path that EPB did, the cost will steadily *drop* over time, especially if they can keep expanding and adding more customers; bandwidth gets cheaper the more of it you buy.

    And by the standards of nearly everywhere else in the country, that is an amazing deal. I'm lucky, and get gigabit for $70, but if I weren't, it's not like $95 would upset me.

    When I first moved here, it was $150/mo for 250MBit, so EPB has been absolutely fantastic. They quadrupled my speed, and cut my bill in half.

  • Dec 29th, 2015 @ 3:49pm

    (untitled comment)

    >Uptake rate has been phenomenal, with eighty-one percent of households signing up for service.

    That's REALLY impressive. EPB in Chattanooga, which offers similar service for a similar price (and is run EXTREMELY well), was pleased as punch when they hit 30% uptake. And they had to advertise *constantly*. Even now, just about six years after their first rollout, they still advertise all the time. (I don't know what their current figures are, but I suspect it must be at least 40% by now.)

    This was their actual plan; I believe they figured it as breakeven at 20%, and hit that fairly quickly. Everything after that has been gravy, and it's working out real well.

    It's interesting, wandering around Chattanooga with a laptop, because it's quite normal to have extremely high-quality free WiFi networks available. It's a slow process, but the Internet is becoming omnipresent here, like air. There's a fundamental scarcity most places, and that's not really true here anymore.

    Anyway, my biggest takeaway was this: just building the network is normally not enough. You've also got to remind people you exist *all the time*. Switching away from other providers is difficult, and most people don't realize how bad their Internet is.

    An 81% uptake is ... dayam. Things must have been BAD in Leverett.

    Oh, and as an aside: muni fiber from the power company works out really well. Their purpose isn't pricing as high as possible, their purpose is serving their residents, so they can pass through the *real* pricing of bandwidth. The only scarcity is peak demand, not total usage. Bandwidth caps are a literal racket.

  • Oct 14th, 2015 @ 3:20pm

    (untitled comment)

    It's worth pointing out that horrible people have rights, too. Maybe it's especially horrible people that have rights, because the only time you really need them is when someone in the government is angry with you, up to and including wanting you dead. And when they're that pissed, they can use their government-granted powers to make you sound like the scummiest scumbag who ever scummed.

    Rights that only exist when you're popular aren't rights, they're revocable privileges, easily removed with a smear job.

    Even if you have no empathy at all for the 'bad guys' (who may not, in fact, be bad at all), protecting everyone's rights is enlightened self-interest, because they protect you, too. In our nasty, nasty system, if you ever need them, you will need them very badly.

  • May 11th, 2015 @ 12:52pm

    Re: In other words

    Yeah, this is precisely what I was logging in to say -- it probably wasn't a hunch, they most likely bugged or tapped this guy somehow.

  • Apr 26th, 2015 @ 3:49am

    (untitled comment)

    I actually have a funny story about this, one that highlights just what scaredy-cats people have turned into over the last generation.

    When I was young, over thirty years ago, I got interested in doing a science fair, and decided that I was going to do electroplating. For whatever reason, I decided it was going to be copper; I have no idea why. So I visited a local electroplating shop to learn how to do it.

    Well, to make a long story short, they sent me home with a sheet of instructions and two big bags of potassium cyanide. I am not kidding. Not even a little bit. I think I was about 13, I walked into that shop, and they sent me home with enough potassium cyanide to probably kill several hundred people. Size memory is a little strange when you're still growing, but I think they would have been roughly quart-size, full of powder. A LOT of it, in other words.

    They were very careful to explain to me that even getting that powder on my skin would kill me. So I was incredibly careful while using it, probably the most cautious I've ever been around anything in my life, except possibly guns. I successfully completed my project, and won a second-place ribbon; it would have been first place, I'm sure, if I'd done a better job on presentation. (I electroplated a bunch of stuff, using pennies as a source, and they came out gorgeous! But I didn't present them well; more time there and I'd have won for sure.)

    Can you even imagine someone doing that in 2015? They'd probably arrest everyone in the shop for doing that today. And a kid wouldn't learn how to do copper electroplating.

    It's actually a lot of fun, and it's surprising how well it comes out. Even very fine detail comes through, unless you put a very thick coat on. You don't need anything complex, either: I did it with a simple power supply, some mason jars, some wire, some target objects, and a bunch of pennies.

    And some potassium cyanide.

  • Nov 25th, 2014 @ 8:34pm

    (untitled comment)

    Heh, yet another thought: as far as T-Mobile is concerned, the nightmare scenario is the family-of-five on the road, with one driver, and four people pulling separate Netflix streams simultaneously.

    That image would keep their network teams up at night. Wired providers shouldn't care in the slightest, but in the wireless business, that is (at present) a very scary idea.

    Four music streams? Not so scary.

  • Nov 25th, 2014 @ 8:23pm

    (untitled comment)

    Oh, and from another angle: I believe the 'free music streaming' is, at its most fundamental, T-Mobile saying, "We have enough bandwidth now to support a live audio stream to a lot of devices at once." What they're probably (rightly) scared of is people using their service for video, which can be something like 25 times as data-intensive as a basic MP3 stream. So they're kind of trying to split the difference, using the data caps to try to keep people from using services they can't truly support yet (like Netflix), while simultaneously letting them use ones they CAN handle, like Spotify.

    In another ten years, it's quite likely that their network will have grown enough that they won't care very much about Netflix anymore: maybe they'll just do away with caps completely, sometime between now and then.

    Again, commit and burstable numbers fix this problem, but if you're being honest in wireless, saying something like '64K commit, 2Mbit burstable' looks very unappealing next to your competitor who's lying his socks off and screaming about FIFTY MEGABITS TO YOUR PHONE, and then adding, in mousetype at the bottom, "until you hit your data cap, fifteen minutes later."

    tl;dr: they don't want to admit that they're lying about bandwidth, like everyone else in the industry, so they're trying to be creative about telling you that using a little bit of data, all the time, is okay, just dear Lord, don't stream Netflix.

  • Nov 25th, 2014 @ 8:13pm

    What they're doing isn't so bad: plus, you're coming from the wrong angle anyway.

    I think you're missing the boat a little bit on this one, Techdirt.

    Having done networking for a long time, one of the fundamental truths of being an ISP is that, for the most part, you're paying for bandwidth. T-Mobile is in that same position; their customers pay them for traffic, and then they buy the capacity to deliver that traffic from other people. (taking, of course, a hefty markup, since they're buying enormous circuits and get a heck of a volume deal.)

    One thing that T-Mobile can do, though, is avoid paying for traffic by peering directly with destination networks. That is, rather than buying transit via Cogent to get bits from Netflix, they can run a wire directly to Netflix, and then carry those bits over their own infrastructure without having to pay anyone else.

    What has net neutrality supporters really upset is when an ISP uses its monopoly control over its customers to extract rent from providers; they're demanding to be paid on both sides for delivering data to customers, when the fundamental idea of the Internet is that you're only supposed to get paid for bandwidth once, and that people are free to buy bandwidth from any provider. In essence, ISPs like AT&T are using their monopoly status to muscle out the competition, taking money away from Cogent, Level 3, and other transit providers, and demanding that it go into their pockets instead. This is profoundly abusive, and it's something to be furious about.

    But just running a link to another network, and avoiding paying for transit? That's just smart engineering. My take is that T-Mobile is saving money by directly peering with different services, and then passing those savings along to the customers. As long as they don't charge the provider anything, it's cool. In fact, it can save money for both sides; Netflix doesn't have to buy bandwidth for T-Mobile customers, and T-Mobile doesn't have to buy bandwidth for Netflix. They have a (significant!) upfront capital cost to buy the routers and run the fiber, and then it's done, and costs almost nothing to run. And all the money being spent goes to other companies entirely, like Cisco and Juniper and so on. There's no conflict of interest, it's just eliminating the middleman, and passing the savings along.

    Now, over the long haul, this could become a problem; T-Mobile might decide it wants to charge Netflix for access to T-Mobile's customers. That, in a word, is garbage. As long as that doesn't happen, as long as the peering agreements don't cost either side anything, other than the capital cost to build the link, then it's not such a big deal.

    And what about small players, without their own network numbers and BGP presence? Most of them are going to be so small that T-Mobile can probably just absorb the streaming cost without hurting too much. If they get big enough to be visible on the usage charts, then they can set up some kind of peering arrangement. If nothing else, they could set up a small remote presence in a colo where T-Mobile has equipment. This isn't hugely expensive, and any service pushing enough traffic to be noticeable on a giant network like that should easily have the money to set something up.

    Further: you're looking at this from the wrong angle, anyway. You're being a little horrified that T-Mobile is charging more for some bits than others. What you should actually be horrified about is the idea of charging per-bit at all. Data caps should not be allowed: ISPs should be required to sell bandwidth as commit speed, and burstable-to speed. This reflects the actual architecture of networking. Transferring bits from place to place is just barely this side of free, until you don't have enough space in your pipes, at which point you have to spend a bunch of money to run a bigger one... and then the run cost goes back to just about zero. Charging per bit is an absolute racket.

    Going with a commit and a burstable-to communicates truthful figures. Commit represents what everyone will get if they're all trying to use their connections at the same time, and will be quite low, but guaranteed to always be delivered, barring major hardware or software failure.

    Burstable-to is what everyone sells now, the headline number, and it's fundamentally a lie. Very few ISPs can actually deliver the bandwidth they're claiming to sell. This is what you should be furious about. Net neutrality is really only an issue because ISPs are under-provisioned. If they weren't allowed to lie anymore, if they were forced to sell only what they could really deliver, then there would be no need for prioritization, because there would be enough bandwidth.

    The whole idea behind caps is to try to monetize a free resource. Bits are so cheap you can think of them as costing nothing; it's not quite true, but it almost is. What actually costs money is peak demand, and that's where the commit and burstable-to figures come in. They represent your slice of the big pipes at the ISP. After you've paid for your slice, it's nobody's business what you do with it. Use it 24x7, or don't use it at all: those pipes cost the same amount either way. There's no marginal cost on usage.... the cost comes from laying in more capacity, and from paying the people to keep it all running. Those costs don't change based on usage.

  • Aug 31st, 2013 @ 7:52am

    Re: Re:

    No, because you know *someone* owns them, and you're depriving that person of the use of their item.

    If, however, you were able to wave your hand and make an exact copy of the electronics you found laying in their house, without changing the original electronics in any way, would that be theft?

    What harm would have been done, and to whom?

    Unauthorized enjoyment is not a crime.

  • Aug 21st, 2013 @ 2:48am

    (untitled comment)

    > Irrational fear is one of the government's biggest allies, whether it's the DHS or a local police force. Claiming bad things will happen keeps bad laws and policies in place. As long as the laws and policies stay on the books and "bad things" fail to happen, officials are constantly "proven" right.

    It's important to note that the same agency making the claims about violent crime going up will also be generating the statistics about violent crime going up.... and we've watched The Wire. A lot of the things they showed were absolutely real, and one of them was the way the police and politicians screw with the crime stats to make themselves look good.

    The same shit they pull to 'reduce' the crime rate can be used in reverse just as easy. So they will be absolutely be 'proven right', no matter what the actual outcome is. They'll make certain.

  • Aug 10th, 2013 @ 10:34am

    (untitled comment)

    But I should also emphasize: storing data in your own home gives you a *lot* more protection than any other method. There's lots more data you could have than just simple email, and if you're storing the data on hardware owned by other people, particularly if it's shared access on a single machine, then your protections against search and seizure are almost nil.

    Keeping your data in your house *probably* means they'll need a warrant to get at it, although with the way they keep redefining things, maybe they'll be able to just hit you with a worm program instead.

  • Aug 10th, 2013 @ 10:32am

    (untitled comment)

    I've been an advocate of 'be your own cloud' for quite awhile, and have in fact been doing it myself for years and years, because I didn't trust the cloud providers. (I thought Google was snooping, not the government.)

    But one thing we truly need to understand, here, is that email is not encrypted, and the NSA has taps all over the Internet. Even if it's encrypted on your local drives, and the NSA would need a warrant to break into your house and read it, it's not encrypted in transit. The NSA can snarf it right off the wire, store it in their huge new data center in Utah, read and analyze every word at their leisure, and track all your acquaintances. If they miss the mail in transit, great, the home cloud will give you a lot of protection, but they probably won't miss it in transit.

    In addition to running our own clouds, we also need to redo the SMTP protocol to include some kind of encryption that's not easily hijacked; I suspect that some kind of distributed 'watch system' for self-signed certificates would probably work fairly well, making it hard for the NSA to do man-in-the-middle attacks without people noticing. It would be nice to also get GPG mail going on a wider basis, but that requires attention by end-users, where SMTP-level protocol encryption will prevent casual snooping.

    Basically, we need to rebuild a lot of the Internet. Not the physical wires, but the protocols; we need to move to encrypted traffic by default, all the time, everywhere, in every protocol except probably games. Hell, even there, we probably want text messages between players encrypted -- witness that one kid who went to jail for months because of game trash-talking.

  • Aug 6th, 2013 @ 7:43am

    Re: Skewed perceptions

    It is always the duty of any government employee to refuse unlawful orders, and to report on illegalities that the government is engaged in, no matter how uncomfortable the government is made as a result.

    Snowden had the courage to live up to his true oath, to protect and defend the Constitution of the United States, not the hired staff presently in charge of the government supporting that Constitution.

More comments from Malor >>