Airport Congestion and Value Pricing

Major airports are often congested which leads to an inefficient allocation of resources. For example, some planes carry low-value packages that are not time-sensitive. Other planes carry high-value, time-sensitive packages. You don’t want to delay the plane carrying the heart for transplant so that a plane carrying 3-day mail can land a bit early. Vernon Smith and co-authors created and tested a combinatorial auction market to allocate airport slots and improve efficiency.  In an auction market when there is congestion the high-value packages can outbid the low-value packages so not every package pays the same freight. Prices in an auction market are also valuable signals telling us where congestion is most severe and expansion most warranted. Revenues from pricing can be used to fund expansion.

Some people, however, are against prioritizing package delivery because they argue that treating every package equally will increase innovation.

Addendum: Don’t take these points as the only or determining considerations but I would like to see more attention given to analysis and less to slogans.


The problem is when the "heart for transplant" airplane is outbid by the "rock star accoutrements" airplane.

To the extent that happens more than sporadically, the inflated prices paid by the indulgent wealthy could be funneled to pay for improved capacity across the entire system.

Exactly, which means when the navel-gazing rock-star bids his gear on board, there's still plenty of room for the heart.

Still, I doubt there's ever going to be a scenario where the final (marginal?) item on board is going to be either of those.

>bids his gear on board

I say there, guitar, come thee onto my flying contrivance!

The whole thing is a lazy and inaccurate metaphor. He leaves out the airport refusing to let planes land that are carrying packages important to the business of their competitors. Or for any other reasons not even necessarily driven by rational thinking that will stifle the efficiency of goods passing through. Weak sauce.

Distribute limited (if that is actually the case) slots with a market pricing mechanism, sure. Allow them to refuse service or achieve that effect through delays for their own firm-level competitive goals when they are the only airport in town and they were / are heavily subsidized with public money -- not so good.

There is no evidence that such blocking, or even limiting of bandwidth, has ever been used for anti-competitive purposes. So what he is "leaving out" is not actually a fact of the situation; it is merely the preferred hypothetical argument of his opponents.

The resolution was a settlement that didn't disclose any details publicly, so what exactly happened isn't clear, but there was a several-year-long dispute between Comcast and Level3 over their network interconnections. Level3 is Netflix's main network provider, and Comcast is both a large consumer ISP and the operator of a competing streaming-video service. There are speculations that the dispute was related to those observations.

"..more attention given to analysis and less to slogans."

You got yourself a good motto for a blog there.

Motto = slogan

True in that one meaning of slogan is "A phrase expressing the aims or nature of an enterprise, organization, or candidate; a motto." But I believe that TC is referring to another one - "A phrase used repeatedly, as in advertising or promotion". But then, he may be just making a play on words.

You should change the scenario so that the airport is owned by one of the airlines. That conflict of interest is at the heart of the debate.

Again, a narrative is being created - don't bring reality into it.

How about the reality that there is no evidence that the alleged problem exists?

Actually the aircraft carrying a heart for transplant would notify ATC and be given priority.

Shh - a silly narrative is being created, and any attempt to bring reality into it is just bad manners. The heart transplant would also be given priority in an auction ruled world, since human beings tend to be motivated by any number of reasons. Including compassion and charity.

If this is supposed to be some kind of "net neutrality" metaphor, it might be worth pointing out that "3 day mail" and "hearts to be transplanted" are poor correlates for data packets, which are virtually identical and travel at the speed of light.

I would also like to see more attention given to analysis....but by people who understand the underlying technical considerations.

So are you saying that bandwidth is not scarce?

Um, not exactly, but this is a perfect example of "not understanding the technical considerations." Bandwidth is just a measurement. Are millimeters scarce? Are gallons scarce?

If you're asking if fiber is scarce, the answer gets more complicated. Yes, fiber is "scarce" in the sense that it's a finite resource and needs someone to put it in the ground. But there is so much fiber already in the ground, much of it not even lit, that we're not even close to capacity, much less getting to the point where bandwidth can be called "scarce."

Exactly. This is just a money grab by companies that aren't going to invest any of the excess monopoly or duopoly rents in infrastructure. It's not needed.

A money grab indeed. These are companies, not naturally inclined to competition, that have actually over-invested in infrastructure, so much so that they had difficulty monetizing it. Some folded, others consolidated. Now they're going to get their rents where they can. It was bad enough when Comcast (NBC) had to compete with your local Ma Bell incumbent and the satellite providers, but now they have to compete with Netflix, Hulu, and Amazon?

Of course they're going to want to get a piece on the back end. If this shyster's not doing it, that shyster would.

The costs are not the fiber in the ground. It is the final distribution to users, the last mile. That is very expensive, and the bandwidth capabilities of that final connection are paid for by that individual user. How dare a service provider act in a way to maximize the return on their investment. Limiting the profit potential for new entrants by rules and regulation guarantees terrible service and entrenched monopolies.

"That is very expensive, and the bandwidth capabilities of that final connection are paid for by that individual user. How dare a service provider act in a way to maximize the return on their investment."

Why are these two sentences together? One might interpret this as saying that the service provider expects a return on the customer's own investment...

Also, if you're worried about barriers to entry, you might be very careful about giving Comcast or an old Bell incumbent the opportunity to charge another toll.

*Sigh* It always seems like the most technical people have the hardest time understanding the relevant scarcities for data transmission.

First of all, bandwidth has multiple meanings. In this context, it means "data transmission rate". A given pipe has a maximum. Sure, it's not scarce in the (strawman) sense of "if I transfer now, I can't transfer later". In that sense, it doesn't make sense to bill for it like water. Point taken.

But it *is* scarce in the sense of "if I use all (say) 10 Gbps on the line for my video call, other people can't send their packets through." So it's close enough that you can see possible related efficiencies that can arise. Let's say that video call is really important, such that I'm willing to pay a large amount for (a data transfer rate equal to) 90% of the pipe, but I have to settle for 1% of it because of all the people torrenting Breaking Bad, who don't even care about latency.

You don't see how that can be an inefficiency resulting from inability to bid for (a kind of) priority?

Personally, my preferred solution would be to lower a user's priority in proportion to their total usage. That incentivizes a reduced load on the network without charging users any more or penalizing them when their requests are at low-congestion times. But no one gives me a serious reply to that because they won't get past "bandwidth isn't scarce herpaderp".

"That incentivizes a reduced load on the network without charging users any more or penalizing them when their requests are at low-congestion times."

I cannot disagree with the logic, but why would we want to "incentivize a reduced load on the network?" Cui bono?

You might be tempted to say that in your scenario it's the video caller is the one who reaps the benefits, but no...not really. The problem that you describe would not exist on a DIA circuit. So again...cui bono?

Seriously, Herb? Who benefits from people reducing their load on the network?

+1 Herb. Three day mail is not a constraint caused by delivery time (packages are stored in the warehouse for two days after arriving in one day), rather its a constraint caused by marketing to get the highest willingness to pay customer to the quick delivery.

In this scenario, online gaming packets can be "hearts to be transplanted" and "youtube clips" can be our "3 day mail". At present, I don't think ISPs are congested but you can argue that's because they make proactive investments in their capacity.

I'm trying to think seriously about this. I'm very much pro-road tolls to deal with congestion, and when I get packages delivered I'm okay with paying less for non-time sensitive deliveries, so why am I against "network discrimination"?

I suppose broadly, it's because it's hard to imagine it being anything beyond rent-seeking by massive, near-monopoly, vertically integrated telecoms/cable companies; increasing capacity is expensive but not nowhere near as constrained as "building new airports".

It'd be more efficient… for telecoms but that's about it. There are fixed costs to running a cable or flooding your house with radiowaves, and so the price will never drop below that.

When we talk about "innovation", the reality is that in a "network discriminated" world will drastically raise the barrier to entry for any bandwidth intensive application. Especially since we have the technical capacity to differentiate between packets carrying video and packets carrying emails. We'll probably end up in a situation not unlike the current cable ecosystem: you have brands that strike deals with the carriers to get placement in one of their bundled packages.

So, suddenly, to reach a majority of Comcast users you'll have to fork over a ton of cash, or convince enough people *before they've tried your app* that you're worth upgrading to the next bandwidth tier. Suddenly, the cost of bringing a startup to market has increased by an order of magnitude.

The incentives in a network discriminating world strike me as maximizing telecom utility over consumer utility.

But at the end of the day it seems we're arguing over the profit margins of Comcast vs the profit margins of Google and - beyond being a tech worker who benefits from the status quo - it's not immediately clear to me that there is a morally obvious winner here.

I tend to prefer solutions that lower barriers to entry for more competition, though. How would we, in economics frameworks, evaluate "the new incentives would constrain growth"?

+1 One still needs to deal with costs of expansion, and congestion at peak periods, but it would seem to me that if the problem is congestion, you charge the consumer directly to bid for capacity during those periods, but you don't impose the cost on the producer trying to send the product through the network. I think carriers would be more hesitant to charge consumers more for congestion, or expansion of capacity, than they would for charging the producer of the product.

+1. This.

Are the airports that are running the auctions also operating their own airlines (which happen to be rather uncompetitive with the independent carriers)? Are the recipients of the packages already paying the airports for the privilege of receiving deliveries? Are the shippers of the packages already paying for departure slots from their airport of origin? No? Then the analogy to ISPs isn't very good. There's a reason why Bjarne Stroustrup said, "Proof by analogy is fraud."

Anyhow, cutting through all the obfuscation, the key point of contention in the net neutrality debate is whether or not ISPs face enough competition to restrain them from abusing their customers. I don't think anybody really knows the answer to that, but there are plenty of reasons to be skeptical. Any critique of net neutrality that doesn't address that question head-on is not worth reading.

It's a key point of contention for sure, but any piece supporting net neutrality which doesn't acknowledge the existence of antitrust laws (and seriously consider their application to these situations) isn't worth reading either.

Fair point. There are remedies if a Comcast won't allow Netflix - at any price - to stream at rates comparable to its On-Demand service.

Possibly relevant:

If anything, US v Microsoft shows why antitrust law is a pretty poor way of handling these sorts of things. The FTC began investigating Microsoft in 1991, and the case didn't go to trial until 1998. The final judgment was entered in 2002, by which time many of the issues under dispute had been overtaken by events. The process was conducted at great cost to taxpayers and Microsoft alike, and the settlement that was finally imposed didn't really do much redress the alleged harm to consumers. I don't think anyone on any side of the issue should consider it a satisfactory outcome.

The thing to keep in mind is that if you are going to rely on antitrust to restrain ISPs, then you've already admitted that competition alone isn't sufficient, and that you want the government to decide the issue. The only things left to argue over are what agency will take the lead and when they will act. You can argue that as bad an option as antitrust is, it's still better than regulation, but I think making that argument convincing will be an uphill fight. In particular, if your strategy is to tell the ISPs, "If the you do XYZ, then we will open an antitrust action," then it's probably better for both consumers and the industry if we just pass a regulation that says, "ISPs can't do XYZ."

For the record, I hate the remedy but for a libertarian site this kind of retarded mechanism is utopia.

Libertarians - let's just call them what they are: crazy people - want to kill off the FDA and have individual consumers sue big pharma when unproven medications cause harm. For example.

The underlying assumption is that a company won't possibly undertake unethical practices because the market will quickly recognize, mobilize, and stop patronizing it, and perhaps even sue it. Monopoly situations aren't accounted for, nor are the costs involved with this kind of process flow.

Arguing the merits of antitrust enforcement with Microsoft as your poster child is a nice rhetorical ploy -- it represents a low-water mark in a number of respects. The previous poster child (and in many ways worse) was US vs. IBM. But the tools are in place for DOJ to police these sorts of arrangements, and doing so outside of a need to get headlines can, and has been, done for quite some time.

The problem with a rule (instead of the threat of enforcement) is that the issues are complex. You'll end up with exactly the same investigations now carried out in a far more amorphous regulatory regime with all sorts of competing concerns and with varying degrees of regulatory capture. Messy as the DOJ system is, it's still considerably better than the FCC; but that's (obviously) just my opinion.

+1 rpl nails the heart of the issue.

I'd be careful about arguing from the metaphor in this case. To put it a different way: there are no transplant hearts on the internet.

Uh, hello - no one on Facebook will care what I'm eating for dinner if they only get to read about it the next morning.

I'll always care, Mitch. We'll always have each other.

'Major airports are often congested which leads to an inefficient allocation of resources.'

And lag times? It isn't as if an Airbus 380 that leaves Tokyo daily is looking for the cheapest airport to land at.

'In an auction market when there is congestion the high-value packages can outbid the low-value packages so not every package pays the same freight.'

Again, the people that paid for a ticket to San Francisco from Tokyo on a scheduled flight, for example, are not looking to bid on where they land.

'Some people, however, are against prioritizing package delivery because they argue that treating every package equally will increase innovation.'

Anyone notice why it was an Airbus 380 that was noted (apart from the fact that I believe the Dreamliner is still not in actual service in any real sense)? There is more than one way to deal with 'congestion,' after all.

Unless, of course, someone intends to profit from that 'congestion,' and has the power to enforce their profit making scheme.

What sort of things should we look at over the next 5-10 years to see which side was 'correct'? Or finding that too simple, what would the two sides consider validating?

For those opposed to net neutrality, I think an uptick in internet infrastructure investment would be something they would hang their hat on. Maybe a decline or slower increase in digital piracy?

For those in favor of net neutrality, I think they would be more story-focused, bringing up examples of specific digital content companies failing or examples of 'worthy' speech/content being impeded or de-prioritized.


I don't think people have a problem paying for bandwidth, which high users already do and telecoms are already allowed to. In your example it would be like if the airport wanted the right to charge the heart transplant more simply because they thought it was more important and they would pay more. The pilots are trying to tell them that it's a plane and it doesn't matter what inside of it to the airport. They're trying to de-commoditize bandwidth and make bandwidth less fungible, that definitely make for a more opaque market with less incentives for innovation in speed or bandwidth.

So profit potential is not an incentive for investing in speed or bandwidth? Or isn't an incentive for competitors to enter a market?

Given the capital costs required, you're likelier to see innovation at the content end than at the physical infrastructure end, and we're quickly approaching the end of useful returns on investment for increased speeds. We could do great things if everyone had 100mbit internet but we'll probably have a hard time justifying 1Gbit internet.

"In your example it would be like if the airport wanted the right to charge the heart transplant more simply because they thought it was more important and they would pay more."

So, the better comparison would be with airlines carrying passenger traffic rather than airports carrying airplane traffic. Passengers are in fact de-commoditized, and airlines do charge business travelers more than tourists "simply because [business travelers value their plane ticket] more [highly] and [will] pay more." The mechanism for the price discrimination is the Saturday stayover, or lack thereof, and the result is that tourists get to fly more cheaply than they would have if everyone were charged the same. Price discrimination for high fixed cost, low marginal cost services usually ends up benefiting those that aren't willing or can't afford to pay high prices, as long as the firm can charge those users low prices without cannibalizing business from high-price users.

So, the idea that people won't be able to innovate because they won't be able to afford the same prices as high-profit firms seems mistaken if the network is uncongested and the network operator is price-discriminating solely for the purpose of maximizing profit. Poor users should be fine as long as network operators can distinguish between poor and rich.

On the other hand, one could argue that some people will be disincentivized to innovate because they will anticipate that, if they do develop profitable technology/services/content/etc., at that point the network operator will start raising their prices, eating away at much of the profit. Megan McArdle wrote a column recently about the business rule: "Never build your house on someone else’s land.” []. People may now be realizing that they are building on someone else's private property.

Poor analogy for many reasons; here are just a few I could think of before coffee:

1) Isn't this more like if UPS and FedEx owned all the airports, and arbitrarily delayed their own/each others' packages to promote their own ground shipping service? (i.e. internet is air travel, TV is ground).

2) What if the airport is a hub? Or, better yet, what if you had to land in four or five different airports on the way to a destination? What's to stop someone from raising tolls at critical junctures to extract large rents? Seems like a mechanism to enable rent seeking to me.

3) Is the airport required to take all comers? What if a given airport didn't allow any flights with turbaned brown men?

4) Who's collecting the rents for this airport auction? In this case, it's private corporations, isn't it? (I'm not normally a rich-get-richer guy, but why are we assigning more rent collection abilities to people that already extract enormous rents?)

5) Can't a rich person who wants to get to Bermuda faster outbid a heart transplant? Should no one get heart transplants around major holidays?

Here's a revolutionary idea: Build more airports. Why should an organization like Federal Express have to compete with commercial air travel? Airports are big patches of paved land; there's no reason it shouldn't cost more than a few hundred thousand dollars, a drop in the bucket for large commercial shipping companies.

Oh but wait the FAA would have a fit. Never mind.

Oh dear, you think modern airport infrastructure costs "hundreds of thousands of dollars"? Sure, they're basically just like glorified parking lots, right?

No. Go look at how much it costs to build a new runway at a modern airport (just the runway. Don't worry about the other bits) then, try again!

Let's be clear. There's not pure net neutrality now. If you want faster bandwidth, all the ISP provides it for a premium. It's just a matter of whether the customer or the sender pays.

For that matter, the underlying data packets have prioritization built in. The internet was built on the concept of varied levels of importance and "net neutrality" is a poorly named concept. The net is not neutral and never has been.

Well, not quite. We can talk about QoS but the difference is your (TCP) packets *still get there*. Under a "network discriminating world" we're imagining, visiting on a low-tier package will simply render a page saying "This premium website has been scrambled. To subscribe, click here and one of our agents will speak to you".

The only real world cases I've seen involved an ISP hosting preferred content on their local servers and providing quicker access to it. I think most of us agree ISP's shouldn't be allowed to block content, but the net neutrality regulations didn't limit themselves to just that particular case.

I want my ISP to give me faster access to some content. As long as they don't deliberately block or slow down access to other content, I don't care about them offering perks for preferred content.

>I want my ISP to give me faster access to some content. As long as they don’t deliberately block or slow down access to other content, I don’t care about them offering perks for preferred content.

Well, they already do this. They're going to prioritize your (latency sensitive) UDP packets while you're sitting on your XBox and deprioritize the TCP packets being sent back and forth between your roommate and Facebook.

It's in their interest to do this because it reduces noticeable congestion on latency sensitive applications.

However, all network neutrality arguments are presumed on the notion that service will either be negated or severely throttled. You can still access Netflix, but at 15kb/s you'll never watch an HD movie in real time. Unless you pay the infrastructure provider to unthrottle it, etc, etc.

This is less nutty if you consider this is already happening with some VoIP providers. A few years back we had stories about how $TELCO_HERE was throttling rival VoIP services to near uselessness while their VoIP service continued to work just fine.

You notice, though, how all the congestion is located around the shops? They want you to spend money. It's similar to the maze of Ikea.

There are really several issues in this post that are all mixed up with each other, so let's disentangle them first.

1. Combinatorial auctions. Great idea, work great when people have preferences for different combinations of items or attributes. That's not the issue.

2. What is the issue: where the seller with market power creates an artificial constraint to differentiate the markets. For most goods, we don't care, but if there is a monopoly element--only one cable provider because of city charters, or limited broadband spectrum allocation, or limited capacity at a publicly owned facility like an airport--we do care.

Both Tyler and Alex make similar pleas on this topic, asking respectively for "less talk about what this “could” mean and more talk about what it likely will mean" and for "more attention given to analysis and less to slogans".

I find this coincidence of nearly identical pleas rather odd. As far as I can tell, as a non-expert, there has been an enormous amount of analysis directed at answering the precisely the question "what it will likely mean", and although I see a good deal of passion, the pieces I've read on the topic have been driven by analysis rather than by slogans.

I'd be very interested to read exactly what Tyler and Alex think about net neutrality, what they think the recent ruling will likely mean, and what they consider to be important and determining considerations.

'I find this coincidence of nearly identical pleas rather odd.'

You clearly have never worked in the sort of world that Profs. Cowen and Tabarrok do, have you?

The only thing that would be worthy of comment is when they actually disagree.

But hey, why trust a former GMU PR employee in terms of telling you what is worth noticing.

Can you help me understand why you spend so much time reading and commenting at Marginal Revolution. Don't get me wrong, I frequently appreciate your comments, but at the same time... For me, for example, reading Z-Mag or Z-Net or whatever they're calling it now is just an infuriating experience. I'd never waste my time reading and commenting on a daily basis over there because it would just be frustrating. So I don't.

But what about you?

I used to be paid by the same sort of people that pay Profs. Cowen and Tabarrok - unlike them, I am ashamed of that period of my professional life.

In other words, this is a place where it is possible to write my unpaid for opinions. A luxury that neither professor seems able to enjoy, even with tenure.

Fair enough. Thanks for being so candid, I do appreciate it.

I find your comments in this thread somewhat cryptic. What are you saying exactly? (No, I have never worked in a university economics department; what am I missing?)

'No, I have never worked in a university economics department; what am I missing?'

I've never worked in a university economics department either. In the time I was at GMU, between the late 70s and early 90s, I worked for private companies, for the Commonwealth of Virginia, and for a center which had nothing to do with economics (apart from making lots and lots of money - Johnson loved having money that was earned by a center's customers as compared to its donors - donors are fickle things with demands, customers tend to be content as long as what they buy is delivered).

The thing is, having a couple of decades experience of GMU, including doing PR, this web site provides a minor opportunity to write what I want, based on what I've seen and heard from any number of previous and current members of the GMU community. After all, I participated in building the facade that was intended to be confused with the reality. Just doing my bit in that cynical smaller steps to a much more prominent university process.

Sometimes what I write does not last long, admittedly, and dancing along that edge does lead to a certain cryptic style. This includes the fact that naming some of those who I used to work with tends to unallowed - even mentioning nationality (which admittedly, can make a person instantly recognizable to anyone familiar with GMU and its growth) is too much.

Of course, what makes this amusing is that the person doing the removal isn't doing it because what is written is false - they are doing it because the information is easily checked using the background of any member of the GMU faculty since the mid-1980s would possess. So I avoid, to the extent possible, providing more context than the minimum necessary - making some statements appear quite unsupported.

A better real world view into this would be to check out the user interface on your cable box.

"Revenues from pricing can be used to fund expansion."

Revenues from pricing can also be frittered away by those with market power to make sure they maintain that market power. Your statement could also be changed to "Revenues from pricing due to stronger and longer pharmaceutical patents can be used to fund more R&D" and retain the same logic.

What about the extra fuel that all airplanes need to carry to be prepared for those "emergencies"? At the beginning of a long flight, a good fraction of fuel is burned just to carry the fuel that's going to be used at the end of the flight. People that cut costs at airlines try to reach the minimum takeoff weight, that means the minimum fuel that is needed to reach the destination safely.

I can only read until page 2 of the article and fuel is not mentioned. I assume transactions are made for medium term contracts not for everyday use, ie passenger flights during the day, cargo during night. I've already experienced that, cheaper EasyJet flights are at 6 in the morning.

This is an idiotic analogy. Markets for shipping, airline, and airport are all competitive - if I order something from San Francisco to be shipped to my home in DC, I can choose FedEx, UPS, DHL, and, yes, USPS; they in turn can elect to fly out of any of the three area airports, SFO, OAK, and SJC, using any of the several dozen airlines that serve them.

Markets for internet service, however, are much more integrated - it's not like my data packet has a choice of tubes. Every single bit that goes in and out of my home has to go through my provider, RCN; if I send you an email, it has to go through your provider, no matter what. ISP markets tend to be non-competitive, with most people having one or two options. And lastly, data packets are just not like actual packages. If I am sending you an object that you need tomorrow, that tends to be different than electronic information, which tends to be instantaneous. Other than instances like high-frequency trading (itself a dubious use of limited social resources) marginal improvements on data speed versus the baseline tend not to lead to dramatically improved outcomes except in the special cases of movie streaming or online gaming. While keeping in mind that these are both vastly less important pursuits than "getting a new heart," even if we stipulate that there could be welfare games by some people, who are already paying for guaranteed service of say 50-100Mbps download speeds being able to pay a premium for their gaming or streaming data (and perhaps receive a discount on their browsing, emailing, etc) we would be simultaneously giving the power to dictate service to a natural monopoly. When it comes to energy, water, gas and the like we call that a "bad idea." Why would it be a good idea with ISPs? Perhaps in a world where technology allows for more competition this could be revised, but there's no reason to suspect that there would be anything but ISPs using their monopoly power to throttle competing services under the status quo.

Seriously. Grow up.

+50. Said much better than I was able to.

Sounds like you're saying curb rights are the more appropriate metaphor here.

I'm baffled by what Alex has posted. Given package delivery by air is handled at night on scheduled flights out of airports where things are coordinated to minimize congestion.

Hearts would not be carried by FedEx, but by courier who would most likely drive because organs are almost entirely allocated regionally.

Further, a heart is a high value package that will incur a $5000 delivery charge if a charge were actually computed, but more likely a heart traveling a long distance would ride by private jet, part of some company's executive travel or plane movement or pilot qualification donated because the marginal cost is low.

But I'm guessing Alex is trying to make some point about the Internet using planes and airports, based on the comments.

So, as a telecom engineer and scientist, let me explain planes and airports and package delivery in telecom terms. It would either be circuit switched, or TDM, depending on the analysis being made.

In circuit switching, moving anything requires scheduling all the resources along the path prior to moving anything.

This is what FedEx and UPS do to a huge degree. Especially when they run the logistics of the shipper, or co-locate with them. Here's how it works: a package source and destination notification are given to UPS/FedEx, with priority and weight. The packages are queued, and the planes and trucks are packed virtually, and all the resources to handle those packages allocated. Then, for each package, the complete circuit is computed and allocated, and resources dispatched. Trucks pickup the packages in the afternoon, they go to the airport, they get sorted and packed on the plane, the plane takes off and lands on a rigid daily schedule, the packages sorted and loaded on trucks, and out for delivery on a precomputed route based on priority and fuel minimization.

I order with "free shipping" which means that for a package created at Amazon's midwest or west logistic center, the package will not be created for days after UPS gets the package info because UPS won't have a truck to carry it east for several days, so it won't tell Amazon it will pick it up. Once a truck has space, the package is picked up, moved to a sorting center, maybe to a second one, then it is on a truck for one to two days going east. And FedEx moves a huge portion by truck even cross country.

The point is, UPS/FedEx do little "on-demand" package handling - they control the acceptance of most packages for shipment so they can allocate the resources for the entire trip.

USPS does on-demand package handling, depending on the law of large numbers to predict the daily load, and then contracts with FedEx and UPS for a fixed quantity of capacity. If the number of packages for a given day exceeds the contract capacity, then congestion "in the network" takes place, and packages get delayed for the untracked packages, which is most.

Oddly, USPS was the first to offer fully tracked delivery, but it was slower because it was tracked and handled only by the most trusted - registered mail, and the tracking added time and effort to create a trail, but never to determine how or when it would travel.

FedEx/UPS fully track packages to fully maximize the utilization of resources - if a truck is delayed, or a package can't be picked up, adjustments are made to each outgoing truck or plane.

Another telecom model is used traditionally for voice, TDM - time division multiplexing. That is the model for take-offs and landings. Fixed slots are laid out and scheduled. A big plane might get a couple of slots, a small plane one or three, depending on speed and what is ahead of it. A landing slot is followed by a takeoff slot if one runway, followed by landing. Planes queue on the ground or air waiting to be given a slot.

In buying gates at an airport, airlines by a share of all the landing and takeoff slots. If conair has 50% of the gates, it basically has 50% of the slots. If the weather is bad and the slots per hour is reduced, it still has 50% of the slots, but that will mean fewer takeoffs/landings an hour for that airline than they might have scheduled based on good weather 365 days a year.

If you want to think about the Internet in a physical world model, just look at the highway network and cars and trucks. The Internet is packet switched. Packets are variable in size, but with upper and lower bounds, like a truck down to a motorcycle.

The packets carry with them their destination, and at each junction, they queue up to take the ramp based on the routing signs that gets them closer to their destination.

At the start and end is a ramp, and the ramps can be big so a lot of traffic can use them, or they can be small. You could think of them being multiple lanes wide, but only if you understand that no passing is allowed.

Each segment of highway has a speed limit which is also the minimum speed. Every car and truck on the Internet goes 70MPH always and exactly. For the local street, everyone goes 30MPH exactly.

If too many cars are getting off the Internet onto a local street at an exit, the cars will crash. The Internet must have wreckers to get rid of the crashed cars quickly, never to be seen again.

Netflix is like the huge sports/gambling complex; it buys ramps off the Interstate that are 5 lanes wide.

Net neutrality is that the complex in northern NJ has lots of ramp capacity on several highways with huge numbers of cars either entering or exiting from all directions on a first come basis.

Verizon sued based on it being Gov Christie. While the Gov promises to serve everyone to the maximum, when he doesn't like NYC's mayor, he closes off half the lanes serving the traffic from NYC so all NJ fans get to the game and half the NY fans never get to the game. Verizon is reserving the right to treat Netflix badly to make Netflix pay for competing with Verizon.

The highways from NY have as much capacity because NY paid NJ in a deal for a lot of capacity, but Verizon is reserving the right to decide who gets access to the highway that NY paid for.

As for reordering and giving priority to some cars all across the US highway system, good luck with that. Its even harder on the Internet. And reordering requires central planners scheduling every car at every stage.

But to restate, Kleinrock designed the Internet packet switching based on cars and highway networks. So think of the data as cars, and the Internet network as the entire NA highway and road network, all single lane, but some faster than others.

Comments for this post are closed