Felix Salmon

File under “unexpected societal benefits of high frequency trading”: it’s doing wonders for building IT infrastructure. Sebastian Anthony and Jeff Hecht both have good overviews of the three — count ‘em — fiber-optic cables being laid deep below the arctic sea floor, all in a $1.5 billion attempt to shave 60 milliseconds, or less, off the amount of time it takes to get digital information from London to Tokyo.

None of this would be possible without global warming, of course:

Each cable will be laid by a pair of ships: an ice breaker that leads the way, and a cable ship. Until now it has been impossible to lay cables in the Arctic Ocean, but the retreat of the Arctic sea ice means that the Northwest Passage is now generally ice-free from August to October; a big enough window that cable can be laid fairly safely.

But global warming alone isn’t enough to make the economics make sense: standard cable ships aren’t rated for icy waters, so polar-rated ships have to be retrofitted for the job instead, at vast expense.

And yet three different companies have managed to make the economics work, even while offering only tiny decreases in latency: according to Arctic Fibre, the speed of the London-Tokyo connection is going to be reduced from 188ms to 168ms, a reduction of just 20ms.

Everybody will benefit from these cables, not least because they will bypass the current “choke points” in the Middle East and in the Luzon Strait between the Philippine and South China seas. But most of us are unwilling to pay for insurance against a ship dragging an anchor in an inopportune location. High-frequency traders, on the other hand, are willing to pay a lot.

The leaders of the project will need to persuade telecommunications companies to buy a piece of the capacity created by the cable. Telecom companies will make that decision largely based on demand from financial companies.

“What we’ve seen is just because you have a diverse path does not mean that you can necessarily sell that capacity for much more than the current market price,” Mauldin said.

I’m a little bit unclear on exactly how these cables are financed, and what the mechanism is whereby telecommunications companies sell ultra-fast connections to financial-services companies. But clearly it’s advanced and predictable enough that the economics make perfect sense for more than one player.

It’s lucky, then, that laying cable under the arctic makes a key financial connection noticeably faster. That cable would serve a very valuable purpose even if was a little bit slower than current connections — but in that event, no one would have any incentive to lay it. Are there any examples of people spending a lot of money to lay big fat pipes which are slower than what already exists but which simply expand the total amount of bandwidth available? I suspect the economics in that case would be much more difficult.

The Justice Department’s official complaint seeking to stop AT&T from taking over T-Mobile minces no words:

T-Mobile in particular – a company with a self-described “challenger brand,” that historically has been a value provider, and that even within the past few months had been developing and deploying “disruptive pricing” plans – places important competitive pressure on its three larger rivals, particularly in terms of pricing, a critically important aspect of competition… unless this acquisition is enjoined, customers of mobile wireless telecommunications services likely will face higher prices, less product variety and innovation, and poorer quality services due to reduced incentives to invest than would exist absent the merger. Because AT&T’s acquisition of T-Mobile likely would substantially lessen competition in violation of Section 7 of the Clayton Act, 15 U.S.C. § 18, the Court should permanently enjoin this acquisition.

One thing which fascinates me is the way in which neither the complaint nor the press release makes any mention of the fact that the proposed deal would give the merged company substantially all of the market in GSM cellphones — the only ones which work in most of the rest of the world. Americans who travel internationally pretty much have to get their cellphone service from one of these two providers — and they’re highly sensitive to exorbitant international roaming fees. Which would almost certainly go up in the event of this merger.

The noises coming from the FCC in the wake of this suit are supportive, with FCC chairman Julius Genachowski saying that he too has “serious concerns about the impact of the proposed transaction on competition.” He adds for good measure that “vibrant competition in wireless services is vital to innovation, investment, economic growth and job creation, and to drive our global leadership in mobile.”

AT&T hasn’t officially given up, but I can’t see it winning this particular fight with the law. This, then, is a good day for the American consumer, not to mention a great day for Sprint and Verizon. AT&T and T-Mobile have both put enormous amounts of management time and shareholders’ money into putting this merger together, all of which will now be for naught. Rather than fight the inevitable, they should go back to fighting each other where it matters: in the marketplace.

I sent AT&T spokesman Mark Siegel some questions this morning:

1. Will you have “rollover megabytes”? If not, why not?

2. Why do the plans have to be chosen ex ante, rather than ex post? Wouldn’t the plans be much more convenient for consumers if they just automatically paid for the Data Pro plan when they went over 200 MB, but paid only for Data Plus if they consumed less than 200 MB?

3. How exactly does data plan switching work? If I’ve consumed 150 MB in a month and switch from Plus to Pro for the rest of the month, do I pay any more than if I had been on Pro all along? What if I’ve consumed 250 MB in that month before making the switch? Do I pay $30 or $25? And what if I switch down from Pro to Plus — is the amount of time I spent on the Pro plan pro-rated, or do I still get the whole month for $15?

4. Here’s a comment I received on my blog:

It’s a patently cyncially priced plan. It’s extremely easy to exceed 200MB if you use your phone to surf the web and use Google Maps on a fairly regular basis, but you’re unlikely to exceed 500MB unless you do data intensive stuff like downloading music and streaming video.

You say you’re all about consumer choice, but it does seem that your choices are clear ones only for (a) Blackberry users who mainly just use email; and (b) heavy users who stream music/video, or who have a 3G iPad. The rest of us — which I think would include most people with an iPhone — are in that unhappy cusp zone around 200 MB where it’s very easy to make the wrong choice. Are these plans specifically designed to make us unhappy?

5. An AT&T representative said here that iPad 3G owners who turn off their $30 unlimited plan will be able to turn it back on again. Is that true? And is it fair for people to characterize the widespread advertising of the unlimited iPad plan as a bait-and-switch, given that it lasted less than 40 days?

Siegel replied:

We don’t have rollover megabytes.

The iPad plans are all prepaid and no-commitment. You pick the plan that works for you. Want to drop it? No problem. Want to pick it up at some other time? Also no problem.

We think that approach is easy and flexible and puts the customer in charge of what they want to do.

On switching plans: Customers can switch between the two new plans easily, even in the middle of the month. They can do so themselves on the Web or by contacting us. In either case, they choose whether to make the jump from DataPlus to DataPro that day, for the next cycle, or backdate to the beginning of the cycle to avoid overage charges. And remember, we give free text message (and email if we have the address) alerts at three usage levels, in addition to all of the other ways customers can monitor usage.

If you buy the iPad before June 7 and want to use the unlimited plan, you can.

So, that’s one question answered, at least: it seems that you can backdate your data plan to the beginning of your billing cycle if you’re switching up from Plus to Pro. (It’s not clear if you can backdate a downswitch from Pro to Plus.)

It’s also pretty clear that if you turn off your unlimited data plan on the iPad, you won’t be able to turn it back on again. But we knew that already, didn’t we.

As for the lack of rollover megabytes, I think that underlines that the underlying business plan here is cynical/evil. AT&T loves to talk about how many people use less than 200 MB of data per month on average — and if they really cared about serving those people, then they would be happy to let them roll over their unused megabytes. But as it is, someone like me who uses less than 200 MB of data per month on average is still probably going to end up subscribing to the Data Pro plan. Which is great for AT&T — an extra $10 a month for them — but is hardly the customer service that Siegel’s making it out to be.

Mark Siegel, the executive director of media relations at AT&T, was upset that I didn’t phone him before posting my blog entry yesterday on his company’s new data plans. He phoned me this morning, and I told him that I assumed the official AT&T press release — which I linked to from my blog — had all the information that the company wanted to release, but that if he wanted to tell me anything else, he was more than welcome to.

And indeed, he did clear up one thing for me. If you’re on the Data Plus plan, that costs you $15 for 200 MB no matter how much data you use. If you use 201 MB in a month, that’s $30; if you use 401 MB, it’s $45, and so on. If you go up to say 1.9 GB in a month, that’s $150 — six times the $25 you would pay to consume the same amount of data on the Data Pro plan.

Is there any point, I asked Siegel, at which AT&T will help a brother out and automatically switch a heavy data user from Data Plus to Data Pro? No, he told me: “Our assumption is that people are intelligent enough to see that they’re going over. People are way smart enough to manage their own usage.”

This puts a large and unnecessary onus onto people with phones, especially phones with WiFi capability. If you only use data-heavy applications like YouTube or Pandora when you’re connected to a WiFi hotspot, you should be fine with the lower data plan — until that fateful day when your WiFi craps out without you noticing, and you rapidly rack up a huge amount of data usage inadvertently.

I also asked Siegel what plan I should use, in the light of my detailed list of how much data I’ve consumed over the past eight months. “In your case it might be a toss-up”, he said, unhelpfully. “It’s up to you to decide.”

Well, that’s one choice I don’t want. Siegel thinks — or at least he told me — that “people don’t want one plan”, and that something along the lines of the plan I proposed in my post ($15 for the first 200 MB, and then $10 per GB thereafter) would constitute trying to fit all of AT&T’s customers into one mold — something he says that, after “months and months of speaking to consumers”, AT&T has learned that they don’t want.

So I asked him who would lose out from that kind of plan. He said: “For somebody who is a relatively light user, a gigabyte would be a much much much higher level of usage than that person would ever engage in, and why would you charge that at all.”

Somehow he forgot that AT&T, with its new plans, is asking anybody who’s likely to go over 200 MB in one month to get charged for two gigabytes of data each month — or face paying more for 201 MB than they would otherwise have to pay for 1.9 GB.

Siegel’s message, which I’m happy to pass on, is this: “One of the things we found is that people don’t want one plan. They don’t want one size that fits all.” Well, I want one plan, and it’s clear to me that AT&T new pricing scheme is deliberately constructed to ensure that a lot of people end up making unnecessary payments — either for using more than 200 MB when they’re on the Data Plus plan, or for using less than 200 MB when they’re on the Data Pro plan.

Of course, if AT&T weren’t evil, it could fix all this at a stroke, and it wouldn’t even need to change the plan pricing. All it would need to do is charge people for data usage ex post, rather than ex ante: if you used less than 200 MB in one month, it would charge you on the Data Plus plan, and if you used more than 200 MB it would charge you on the Data Pro plan.

But that would be far too easy for AT&T’s customers, and it would deprive AT&T of all that extra revenue from people who guess their data-usage needs incorrectly. Obviously AT&T prefers to make life harder for its customers, if that’s going to give it a little bit more money.

On the London Underground, you don’t need to decide whether it makes more sense to buy an individual ticket or to buy a daily or a weekly or a monthly pass. With the Oyster card, you just tap in and tap out around the system, and it charges you whatever’s cheapest. You only make one journey? You only get charged for one journey. The minute that your journeys in one day add up to more than the daily-pass rate, you get charged the daily-pass rate, and no more. Similarly for your journeys in one week, with the weekly pass. And so on. Really, there’s only one plan, and there’s no way to get inadvertently ripped off.

When AT&T decided to abolish unlimited data usage on its smartphones, that’s the kind of of plan it should have implemented. Instead, it went the evil route, and it’s forcing its current customers to make one of three different choices, based on limited information. Whatever they choose is quite likely to be the wrong choice, and AT&T will chortle as it collects all that extra money which its customers didn’t need to pay.

The first choice is known as Data Plus, and gives 200 MB of data for $15. If you go over the 200 MB cap, you pay another $15 for another 200 MB. If you go over that cap, it’s not clear what happens, but you’ve already paid $30 and will certainly be asked to pay more.

The second choice, Data Pro, gives you 2 GB of data for $25, and then $10 per GB thereafter.

And the third choice is to stay grandfathered in to the current plan, which is $30 per month for unlimited data usage.

You can switch as much as you like between Plus and Pro, but once you leave the unlimited plan, you’ve left forever; you can’t go back.

AT&T is good at disingenuous statements like this:

Currently, 65 percent of AT&T smartphone customers use less than 200 MB of data per month on average.

This is disingenuous on two levels. First, as John Gruber points out, it carefully talks about “smartphones” rather than iPhones: the number for iPhone users is surely significantly lower.

But second, just because you’re using less than 200 MB of data on average doesn’t mean that you should necessarily choose Data Plus. I’ve just had a look over my most recent iPhone bills, and here’s my monthly data usage over the past 8 months: 202, 120, 160, 143, 89, 39, 333, 287. On average, I’m using 172 MB of data per month, and even with the overage charges I would have been better off with Plus rather than Pro. But for the past couple of months I’ve been significantly over 200 MB, and would be better off with Pro rather than Plus. And then, if I get the new iPhone 4G, is that going to raise my data consumption? Who knows.

At least with the subway you’re in control of how much you use it. With data usage on a phone, it often comes down to questions consumers can’t be expected to understand: how much data does say Google Maps use? And, more generally, if the AT&T network is good, and doesn’t time out on a regular basis, you’re going to use it more. And consumers can’t reasonably predict how good the AT&T network is going to be next month.

AT&T could easily have saved consumers all the trouble of having to try to predict their next month’s data usage by having a single plan: $15 for the first 200 MB, say, and then $10 per GB thereafter. They didn’t, because they’re looking forward to getting $30 per month from people exceeding 200 MB of data but who use nowhere near the 2 GB that “Pro” users get for $25. That’s where AT&T is evil, even if you think (contra Jeff Jarvis) that it makes sense to abolish unlimited plans.

For reasons that I don’t fully understand, Jim Surowiecki is not a fan of a-la-carte cable pricing, where you just pay for the channels you want, and not for the channels you don’t want. My views on the matter haven’t changed since 2007, but one big thing has: broadcast TV is now digital, which means that cable companies are extremely constrained as to how much they could charge for network TV in an a-la-carte world. As a result, it seems obvious to me that consumers would likely save a lot of money if they paid only for the channels they watched. (The one possible exception? Sports fans.)

Surowiecki says that consumers wouldn’t benefit much from unbundling cable channels, while cable providers and cable networks would both get hurt. He says that a recent paper “estimated the best-case gain to consumers at thirty-five cents a month”, but if you read the paper in question, it doesn’t model a-la-carte pricing at all, but rather scheme where you get a choice of seven different “themed tiers” of programming on top of a $29-a-month basic bill. What’s more, the “best-case gain to consumers” is calculated not as the amount of money consumers save, but rather “the change in expected bundle utility”, taking into account how much “bundle utility” they lose out on by not watching channels they currently watch but which become too expensive under the new system.

There will surely be a lot of unforeseeable consequences to moving to an a-la-carte system: there’s no particular reason, for instance, that cable providers should receive all of the monthly fees, while cable networks receive substantially all of the advertising revenue. My guess is that in an a-la-carte world, both would be shared much more equally.

But as Surowiecki hints, there’s an extra possible step here. If you move along the spectrum from full bundling, like we have right now, to “themed tiers”, to a-la-carte, and keep on going, you end up at a simple metered system not unlike that which the NYT is mulling. Give everybody every channel, let them watch whatever they want to watch, and then bill them for whatever they consumed. Mass-market channels with mass-market advertisers would be completely free — as will smaller channels seeking to build an audience. High-end HBO dramas and big-time sporting events would be quite expensive.

I think the problem with that kind of system is the same as the problem with any micropayments system: people don’t like the psychic cost of paying even a small amount for anything, and would much rather just make one big payment and have it over and done with. As Surowiecki says, bundling “eliminates the problem of fretting about small expenditures”. TV is something people have on in the background: they don’t want to worry about running up a bill for that kind of activity.

The ideal world, I think, would be one where consumers got to choose. You only want to watch the Daily Show, 30 Rock, and breaking news? Go for metered pricing. You want HGTV, Bravo, HBO, Comedy Central, and ESPN? Go ahead and buy them. You want full access to the whole menu of hundreds of channels you can watch at any time at zero marginal cost, just like you have right now? Fine, you can keep what you’ve got.

That world would be great for consumers, but there would be a lot of adverse selection: people currently paying large sums for TV they barely watch would immediately trade down to a cheaper option, and the cable providers would lose that extra subscription revenue in perpetuity. Which means that unless and until it gets mandated by the FCC, it ain’t gonna happen. I’m not holding my breath.

Andrew Vanacore has a long and sometimes confusing overview of the state of play in the television industry, which concentrates on the possibility that one or more networks might convert into cable channels sooner or later. But there seem to be lacunae in the story — not least the large number of cable channels which pay for the privilege of being featured in the cable-TV lineup, rather than being paid by the cable companies.

Also missing is any indication of the effects of the move to digital broadcasting. Reader William Wang notes that this vastly increases both the quantity and the quality of free-to-air television stations, which, combined with Hulu and Boxee and YouTube and all the other free sources of TV, should put a lot of pressure on cable companies who have historically had local monopolies and the ability to raise their prices every year with impunity.

The networks are increasingly fighting with the cable operators, now that the likes of Rupert Murdoch have decided that network TV, just like newspapers, is something which should have more than one revenue stream. Now free-to-air digital broadcast TV does not, of, course, come with any kind of monthly subscription stream from a cable operator desperate to still be able to serve up American Idol to its loyal customers. But the networks still reach vastly more viewers than any cable channel, and so if they started adding their in-house cable channels, like Fox News, to the digital broadcast spectrum, they might be able to get a significant bump in viewership.

I still think that by far the best outcome for most constituencies would be a la carte pricing, where viewers — rather than cable companies — decide what’s worth paying for. Every television station on cable could then charge as much as it liked, with an eye to maximizing the sum of subscription revenues and advertising revenues. But what’s clear is that we’re moving to a world where the number of options is multiplying, and corporate strategy is going to get extremely complex extremely quickly.

The latest round of fights between producers and distributors smells like the dying gasp of a 20th-century TV business model to me, with essentially only two sides in the game. The consumer is left out in the cold, shivering as cable bills rise much faster than inflation. Pretty soon, the consumer is going to have a lot more power, and that’s going to change the game in profound and fundamental ways.

Is AT&T’s inability to provide decent wireless broadband in tech-savvy cities like New York and San Francisco a simple matter of physics? Phorgy says yes:

At low frequencies, radio waves are kind of like molasses. They can ooze around corners and through buildings…

As frequencies increase, the waves start acting more like laser beams. They no longer ooze around corners. You start to get “shadows” or dead spots with no signal. It becomes more difficult for the signals to penetrate walls etc. These problems get worse the higher you go in frequency…

So we have two extremes: low frequency molasses waves and high frequency laser beams. As bandwidth demands increase, we begin moving the dial away from molasses (where we have good wireless signals) to laser beams (where we have dark spots, shadows, with no signal, etc).

There are many clever modulation tricks that delay the inevitable, but the basic rule is that you cannot defeat Heisenberg.

The fundamental insight here is true: smartphones operate at very high frequencies, sometimes above 2.5 GHz. And covering a city with wireless broadband at those frequencies is hard. But after a brief email exchange with Ultimi Barbarorum’s Baruch, who knows much more about this kind of thing than I do, I’m far from convinced that what we’re seeing right now with AT&T has anything to do with the hard physical limits that Phorgy is talking about.

For one thing, Phorgy’s limit of 1,000mps in total for a few city blocks is I think far higher than anything AT&T is currently able to provide. With what Baruch calls “compression, prioritisation, all that level 4-7 stuff you can do at the packet level” (don’t ask me), you can serve a lot of people with that kind of bandwidth.

But more to the point, we already have a real-world counterexample which pretty much disproves any thesis that AT&T is bumping up against Heisenbergian limits. Phorgy’s saying that the problem with New York and San Francisco is that (a) they are very high density, complete with skyscraper canyons and the like; and (b) they have lots of people all trying to use 3G cellphones at once. But there’s a city with even higher density than either of those two, with even more people using 3G cellphones, and which has no signal-quality problems at all.

Tokyo is proof positive that this can be done.

The bottleneck is money, not Heisenberg. Building a Japan-style 3G network is extremely expensive: it involves lots of high-frequency and low-frequency base stations and base station relays and femtocells and backhaul capacity and IT investment and other things I don’t pretend to understand.

And of course this kind of investment doesn’t scale. If AT&T puts enormous of money into solving the problems of New York and San Francisco, they can use similar techniques in, um, Chicago. But nowhere else in the US has that kind of density — in contrast to Japan, where density issues are endemic.

If I were an AT&T shareholder, then, I’m not sure how much money I’d want my company to spend on beefing up 3G wireless in New York and San Francisco, especially when there’s little obvious return on such an enormous investment. Sometimes you make more money with cheaper unhappy customers than with more expensive happy customers. And this could well be one of those times.

Update: Some good comments, especially from Mark. Meanwhile, Paul Krugman asks whether central Tokyo might not be such a great counterexample after all: is it maybe less dense than Manhattan? Probably depends on how you define the two areas. But there are other cities which serve just as well: Hong Kong, for example.

Olga Kharif speculates that AT&T might release a crippled iPhone later this month:

The exclusive U.S. iPhone service provider is considering cutting the price of its monthly service package or offering a range of lower-priced plans, say people with knowledge of the company’s thinking. One plan that could be introduced as early as late May would include limited data access at a $10 monthly reduction, the people say.

This is a stunningly bad idea. For one thing, $10 a month is not very much money, considering that the cheapest iPhone plan is $70 per month excluding text messages, and unlimited texts (which aren’t included in the unlimited data plan) cost an extra $20 a month.

More to the point, unlimited data is intrinsic to how the iPhone works. Everything from visual voicemail to thousands of different apps, including very data-heavy apps like Google Maps, relies on the fact that the marginal cost of data is zero. If you start limiting data access, you’re essentially limiting the use of the phone itself, and it becomes something to be mistrusted rather than something to be loved. If I press the Mail button by mistake, how much data will I inadvertently use? What happens if I think I’m happily surfing on WiFi, but then for some reason get booted onto the cellular network?

The iPhone was the first phone to be fully integrated into the internet; unintegrating it by introducing a limited-data plan would be a horribly retrograde step, especially when lots of other sexy new phones, including the Palm Pre, are about to be introduced.

Incidentally, at the end of that NYT article, we find this:

“Phones don’t stand the test of time,” Mr. Donovan said. “I look at my personal handset museum, and the coolest thing I had in my pocket eight years ago is laughable.” When it comes to phones, he added, “there are no ‘Citizen Kanes’ out there.”

Not true.