Book Recommendation: Internet Architecture and Innovation
On the heels of its paperback release, I would like to recommend a book that is required reading in our shop, Internet Architecture and Innovation by Barbara van Schewick. I've mentioned Barbara before on this blog. She is a Law School Professor at Stanford and has written extensively on this important topic.
I am a fervent believer that everything that is great about the Internet emanates from its underlying architecture. That's where Barbara is coming from too, which is probably why we like her and her work so much at USV.
In this book, Barbara argues that the massive amount of innovation that has been made on top of the Internet is directly attributable to its architecture and that there are some parts of the architecture, most notably the last mile in wired and mobile, that are problematic and that do not benefit from the hypercompetitive nature of the open Internet.
This is an important piece of policy work and anyone who cares about the Internet ought to give it a read.
Comments (Archived):
Looks really interesting, thank you for posting it. Will download in the next day or two, once I get my new credit card, with the old one cancelled because of fraud, and cannot download except with default card. On the other hand, card networks have made huge strides in fraud prevention!
For small online purchases I buy Amex gift cards — they do have a small activation fee, but they make great burn cards.
I had an idea over a decade ago to embed something like RSA SecurID numbers in credit/debit cards, so they become one-time use. Essentially, without the physical card, no way to do fraud. I even spoke to RSA – knew the product director back then, cannot remember his name for the life of me – and the cost of manufacturing was ~$1.00, but they were only producing tens of thousands a year; get it up to millions and the cost dropped to pennies to $0.10, IIRC, so manufacturing costs became a non-issue.Everyone was focusing on fraud detection (still do); I wanted to make fraud almost impossible in the first place.Somewhere I have all of the research and groundwork I did back then.Cannot recall what stumbling block I hit. I think I decided that the market wasn’t ready as most CC purchases were still “swipe now, transmit later,” very little was online.
Clever. Mastercard and Discover? do single use numbers IIRC.
Could be. If so, they went at least partially down my path.
Re: last mile. Are the telco’s, city planners or policy makers to blame? Or all of them. There was something called Wi-Max but it’s being phased out, and it was still controlled by the telco’s. Can we buy our broadband from a non-telco or cable-co?
Look to Coase Theorem for the answer. What we have now is regulated oligopoly, or monopsony.
monopsony? Isn’t that the opposite, where there are multiple sellers but only one buyer, as opposed to monopoly, multiple buyers and one seller?
you know, I made a mistake in logic. you are correct. oligopoly is the correct term. it is not monopoly, although the carriers have virtual monopolistic powers because of regulation and high barriers to entry.
I would lay it at city planners and policy makers. In most cases where you can bring from a miserable cable co or even more miserable telco, the regs gave a protected monopoly. To my mind, this is why Net Neutrality even became an issue. If there were sufficient competition, it would happen on its own.
Dang, it’s nice to be back in the AVC.com conversations after a few weeks.
missed you
I must not be causing enough trouble… 😉
That’s what I noticed. 😉
missed you, will email you later, life is still very hectic for me from the storm.
“I would lay it at city planners and policy makers”Aren’t they mostly just corporate-money puppets at this point?
Well, yes and no. They are definitely beholden to corporate money, union money, private wealthy money, bundlers from lots of people like me and you, etc. as well as toeing the line for the media who can make or break either way. But in the end, they still need to get elected, right?As messy as money is, at least it gets spent by everybody in all directions.Does this make me a hopelessly optimistic cynic? LOL!
city planners are puppets of public unions first, then corporations.
Unions don’t have the power to influence much these days.Even at the municipal level developers have all the money and power!Looking over your shoulder for the now defunct Union boogyman is now IMHO pointless and irrelevant.
that would be cool – a coop of broadband.
Does look very interesting but at $24 it’s over my kindle rental threshold. I did download a few of her papers though.Btw, in your link to her twitter you’ve double ‘vanned’.
well above my standard kindle rates as well. there are some videos of her lectures on youtube which i think i’ll check out: http://www.youtube.com/watc…
Good one, cheers!
her doctoral dissertation is also on Lessig’s website:http://www.lessig.org/blog/…
nice find, score one for the frugal ($24 for a book has to be phenomenal for me to grab it )
highest i ever paid for a kindle book was just under $19. made me cringe but i couldnt find an adequate substitute. book ended up being not too good, i didnt even finish it.
I’ve got a dozen untouched books in my kindle queue. Between coding posts/tech docs, and fun sci-fi books my “inspiration/startup” set has been getting dusty
thanks for posting these links
On this occasion the kindle price is substantially lower than the hard and paper books.I don’t get how Amazon prices its inventory.I remember when I worked in a medical bookshop attached to a medical school. Some of the books were priced $x,xxx retail …each.
It’s the same price as the paperback.Edit: Actually, if you sign out of Prime it’s $10 cheaper. Lame. cc @kidmercury:disqus
http://www.amazon.com/Inter…this book?
Yes, that book. The price depends whether you’re logged in or not, and if you are logged in whether you have Prime or not.
I don’t have prime. Is a book sold through a prime account cheaper than a ‘secondary’ account?
In this case it’s $10 more expensive. See my comment above.
oh, not for me then 🙂
dynamic pricing is criminal
i’m seeing the same price when i’m logged in or logged out. maybe it’s a geo thing? or some type of a deeper personalization beyond prime/no prime? i do have prime when i’m logged in.logged out screenshot: http://www.informedtrades.c…logged in screenshot:http://www.informedtrades.c…amazon has taken some heat for its dynamic pricing before. i hope they are not charging prime customers more. that’s not the kind of thing that makes me an amazon fanboy.
Interesting. I did a little experiment (all with a completely fresh browser):German IP & Swedish IP:Logged out – $20Logged in (std) – $20Logged in (Prime) – $24Australian IP:Logged out – $15Logged in (std) – $15Logged in (Prime) – $24US IP:Logged out – $15Logged in (std) – $15Logged in (Prime) – $24I’m very unimpressed.
i’m impressed with your experiment! although quite saddened by the results. i feel betrayed by amazon. prime customers are supposed to be treated better. i buy everything off amazon, but perhaps that will change if i see this type of stuff.
It really pissed me off so I did a little searching and discovered this isn’t an isolated case.Another interesting alleged tactic (I haven’t tested it) is for the same URI to deliver a Prime customer to a 3rd party seller product, and a logged out customer to the Amazon product, thereby “tricking” the Prime member in to paying shipping charges.All this cloak and dagger dynamic pricing has turn an impulse buyer (muggins here) into a paranoid price watcher.
what? that is lame! And unfair!
any favorite papers?
I haven’t read them yet 🙂
Have not read the book but if the thesis is that internet access should be thought of like roads, water & power, I whole heartedly agree.Highly innovative internet areas had state owned telcos or oligarchy telcos. Its foundational, to use the Houston Rockets word of the week.
Information network fundamentally different from other networks, although there are similarities at various (horizontal) layers and vertical boundary points. Almost all of them (including roadways) are of a 1-way nature, whereas much of communications is 2-way; either real-time or store and forward. Therefore the economics and architecture are fundamentally different.
Similar impact on society? I say yes.
Similar impact? Not really. In a 2-way model costs increase linearly while value grows geometrically. That’s why competitive networks can achieve universal service as long as there is government mandated or market fostered equal access. Typically it is because of local regulations that restrict access that we have monopolies.Further, there are significant differences in scale between information and other networks based on supply/demand assumptions at every layer and boundary point. That’s why lower, middle and upper clearinghouse models will work.
I am thinking user impact, but I get your point.
what is “the internet’s underlying architecture”? the current architecture of the web is one which ICANN controls the root DNS zone. we see other countries like china and iran breaking away and basically rolling their own internet. mobile is already like this, where googlenet is different from amazonnet and applenet. i havent read the book, as i have an ideological problem with the authors who charge the same amount for a physical copy as for the kindle copy — i won’t do it unless the material is so good and i’m desperate for it — but calls for regulation, which is basically what this is, always ignore the full cost of regulation (to taxpayers and in the sense of opportunity cost — i.e. lost business models) and the possiblity of unitended consequences (new forms of captive regulators). the author seems to believe the current architecture should persist, which is an argument i disagree with; i think it is evolving and will continue to evolve to a world with an increasing number of internets that ultimately form federations to deal with cross-internet issues. the federation emerging around android (google, amazon, barnes and noble, and best buy) offers a glimpse of what, in my opinion, is to come.
I tend to agree with you when it comes to regulation, and law professors are notorious for being convinced that the solution to everything is regulation – nature of the beast and their education, as well as their incentives and self-interests.But I might actually learn something that will shift my views, or at least give me food for thoughts, in an area that matters very greatly to me (and modern developed society at large). So, yeah, I will read it, and either (a) learn something and shift views; or (b) disagree but at least understand the other position.Hopefully, a little bit of both.
also look at the direction governments want to take the internet: http://www.gmanetwork.com/n…the more control they have, the more they will use it to exercise free speech regulation and all that stuff. the UN takeover of the internet is related to the US debt problem, as the US will increasingly be asked to give up stuff/power in exchange for debt forgiveness. next step in the banker debt scam.
Isn’t this one of those “walking the razors edge” kind of affairs where we need to balance off bureaucratic government interference against corporate monopoly interference especially given wireless bandwidth and last mile access constraints?
i dont see it that way. google is getting into the ISP business. if it’s such a big problem, why aren’t others?
People always work with their incentives. The more outrageous things people can say, the more some ideas that are better for the people than incumbent politicians and bureaucratic regulators will get through and threaten their privilege.Governments (which are made up of real, frail, incentive-driven humans, just like every employee in every corporation, no more or less angelic or demonic than them, just human) always seek to increase their power. That is why we have the checks and balances system we do.
Did you mean “to exorcise” free speech? 🙂 Happy Halloween.
lol quite a timely pun! 🙂
What do you think ICANN should be doing?
nothing. i think they are an increasingly obsolete organization impeding growth of the internet. their role of managing the DNS address book can be handled by Google for GoogleNet, Amazon for AmazonNet, etc.
“Barbara argues that the massive amount of innovation that has been made on top of the Internet is directly applicable to its architecture”Do you mean “directly attributable?”
Yes. Thanks. Will fix.
You’ve still got the double “van” in Barbara’s name.
One for the name, one for the mini-van for the kids? 🙂
Shit. I’m clearly a bit out of sorts today
Understandable, I’d say.
If you’re implying that the “last mile” cannot be done via the private sector, we cannot assume it will be accomplished via the public. The laying of improved bandwidth to the more rural America did not move according to the timeline promised. Add to that, the bickering/cronyism where the big gets the fruit and the small are left behind.Pushing toward the best transfer of Knowledge/Goods/Currency should be at the top of priorities. We just need the team of leaders who have the moral code to make it happen for everyone.
Does anyone remember the term “statesman”?
I probably understood wrong. @takingpitches:disqus has a good comment post in this batch.In the realm of equaling the strength of broadband between land and air, it will take more than a statesman for it is a matter of someone who can address the issue identifying the need, cost and not be afraid to stand by the proposed solution. IOW, the right combination of Macro/Micromanager.
I have a long trip coming up with some very ugly layovers and primitive technologically so this may be a good one.Thanks for the tip.
Barbara discusses the importance of innovators not having to get permission from the providers to innovate, which leads to more innovation.This is right on. We lived it. In the late 1990s, it was a hopeless task when we tried to negotiate with mobile providers to get access to their nascent platforms. Now, you innovate on Android or iOS and, while you need some permission from the owners of those platforms, it’s orders of magnitudes easier than it was dealing with the mobile providers.So net neutrality – not discriminating between data packets – is important because it prevents the last mile providers from regaining control. For innovation’s sake — as she says — it’s important that the demand side for new applications are responding to customer preferences (i.e. actual customer needs) and not to wireless/telco/MSO preferences (i.e., more revenue, protecting their other businesses).The best way to achieve net neutrality would be to completely rethink how we think about spectrum and let innovation happen there through expanding white-space. I and others have commented on this before here.But that’s going to take a while. Until then, you need net neutrality through regulatory fiat to allow innovators to innovate without asking permission and to maintain the raw demand side response to those innovations from customers.But we shouldn’t also be blind to other ways through which wireless/telco/MSOs are trying to take supplant the robust demand sifting function that happens through customers able to experience and judge new applications.The change in pricing structure from flat-rate to metered service is a critical threat today. If customers are hesitant to try new applications because of worry about burning more bytes, this hinders the powerful demand-side function that supports the innovation ecosystem which Barbara discusses. It’s innovators having to seek permission from mobile providers instead of customers again.– straight back to the 1990s.So while it may be sufficient from a net neutrality perspective to think of bytes as bytes, from a demand-side broadband pricing perspective, we need to think of access for individuals instead of bytes. This is inherent in the pricing structure, metered pricing versus access (bucket) pricing.There is precedent for this.We used to think about local phone service from an access perspective (the FCC would ensure that local phone monopolies offered a flat-rate local plan among the telephone options because local phone access was critical and because local phone monopolies were benefiting from a monopoly franchise.) You can deal with concerns about broadband investment by setting the price of the access plan at a level that involves investment.There was also a lot of work among OECD telecom regulators in the late 1990s that concluded that flat rate pricing for ISPs was necessary for the growth of the Internet and e-commerce. That seems to have been forgotten by the current regulators as the wireless companies retreat from flat-rate broadband pricing.In addition to serious spectrum reform, the FCC needs to step in and ensure that both terrestrial and wireless broadband access have flat-rate options available. That is even more critical than local phone access used to be because not is it just as important to local users (think of our information sources during Sandy) but it also supports a vibrant innovation ecosystem that is the brightest spot in our economy.
Very good write up. On the FCC side, a flat rate would do a lot. To fill in the empty spaces, you can have a level fee from the carriers joined with a secondary based on % of phone/tab market… for that matter, you can include the landline cable carriers.On the innovation side, the creation of the truly optimal OS that is further out than what the current players are doing would truly change the equation. It is an expensive gamble, yet putting forth the money utilizing the truly best players can achieve it.
spectrum deregulation is most important. doesn’t get enough attention even from within our own industry.
Deregulation is a 2-edged sword. It can lead to over concentration and monopoly control, or if the government says that it is an inherently shared “public” medium then some measure of open access, be it private (safety), partly private (cellular) or purely public (wifi) is good. That “open access” is “regulation”.
there is no scarcity and thus no risk of monpolization; regulating spectrum is like regulating color. government creates the illusion of scarcity by saying that only certain entities can use certain parts of the spectrum. it would be like if government said only people who paid certain license fees were allowed to wear the color red. there is no need to protect the color red from monpolization, there is more than enough for everyone — it is non-scarce and non-rival.
Wireless ecosystems are fragile. And it is very easy for one user to overpower another. Wifi works because it is government mandated “nano-cellular” via power limits. To put it into lay terms the person with the biggest, most aggressive cows will win in the commons in a purely deregulated model.I agree that regulation should be more flexible and always favor open and competitive solutions (I was a big fan of CDMA and believe our multi-protocol, competitive approach in the 1990s lead to our smartphone and LTE dominance, in part) over rigid, monopoly approaches. But “deregulation” is a term that has lead to higher, not lower, prices.
I completely agree that you cannot discriminate about what the bytes are. It doesn’t matter what I am using my computer for wasting time on a flight across country or coming up with the greatest algorithm known to humankind, that is up to me.Saying that you can’t meter is not realistic. There are real costs to delivering bandwidth. If it becomes cheap enough like voice that it doesn’t matter to meter great. However, that just is not the case now. Saying it gives the providers power because they can punch right through that argument and then start metering on the application.
I’m curious to understand why you say it’s not realistic to have flat-rate plans.T and VZ can earn their cost of capital by setting the price for the bucket rate at a level that allows them to meet their investment requirements.Also, worth pointing out, that even as it stands today and has been the case for a long while, T and VZ earn enormous levels of gross and net income. Not surprising given it’s close to a duopoly today with two marginal national competitors. Despite what they say, they have the money to invest even under the flat rate data plans, which was industry default until recently.
Look a router can only handle so much traffic, same for a cell tower, same for the backhaul If I wanted to flood one of those points, there would only be three options. Meter me, or slow me down, or give everybody else crappy service. Those are the cold hard facts.Just saying flat rate, flat rate, doesn’t make it true. It is no different than metering electricity. If I decided to setup a million space heaters in my house and there was no meter and no throttle (both of which exist and nobody complains about). I would take down the grid for my town.Does that sound reasonable? Hell no. Instead I pay by the kilowatt hour and I am limited to 200amp service. Its no different than what happens if you have your servers in a datacenter. You pay by the size of the pipe and how much peak you use.Why people expect this I don’t know. Frankly I’d rather have it metered because I don’t feel like subsidizing the person that has setup a full time streaming business on the same access point I use.On the other hand if the electric company told me I had to pay more for electricity when I was using a computer versus a light bulb I would be incensed. That is what the phone companies would like to do and it is bs. But when you use a silly argument they can use that as a Trojan horse to get what they want.As for how much they make. Would you like to have made that investment? What would you want in return if you did. Its profit margin is 2.7% Its return on assets is 4.1% think Fred would be happy with that?? If you have a specific argument make it but don’t just say they make an enormous amount of gross and net income. Their stock price which is publicly traded just doesn’t support your argument.
Fair points.I would be more inclined to believe that the move to metered pricing was about network congestion management if it was more granular, e.g., charged more for bandwidth at certain “peak” times. While network congestion is a real issue, the change in pricing structure — without a lot of oversight and transparency — is perhaps also a way to take advantage of market structure.I still maintain some bucket pricing option is better for the innovation environment more generally, but take your point that policy would have to account for network hogs and real congestion issues.
History repeats itself: http://en.wikipedia.org/wik…Initially there were no electric meters and when they went in people did not like them. They charged by the lamp not by how much you used it.Just like electricity yes you can charge less for off peak times. There are two arguments. Letting you do what you want with electricity and charging for electricity.I am in complete agreement when I see people say I do not want VZ dictating what I can do with their dumb pipe. http://en.wikipedia.org/wik… Right there with you.But then when people make the argument we should make VZ provide a person unlimited access you lose me and by combining it confuses the issue, and therefore lets VZ obfuscate the fact that they feel they should get some value of the transactions that happen on their network, versus being content that as a duopoly they just get to deliver the bits.
You’d also cook yourself like a lobster – whereas with data it isn’t quite the same. I think in general metering is the way to go past a certain threshold of certain sites.
It’s important not to confuse 1-way vs 2-way networking models (or 1.5 way; aka store and forward for that matter) and also keep in mind distinctions between stock vs flow rates and pricing. Lastly, the marginal costs at every layer across every boundary point in the information or communications stack vary dramatically depending on context, application and market segment.The industry is beholden to flat, two-dimensional models that do not allow for this complexity to be understood after the fact, let alone a priori. That’s why VC funding is down 70%. So we (academics, regulators, capitalists and trade management) support monopolies that default to average pricing at rates that are 20-150x more expensive than they should be based on reasonable marginal cost assumptions.
I respect your opinions but when I see the market cap of all telecom stocks combined is less than Apple I wonder.I think that the Federal Government has allowed too much consolidation across all industries: Banking, Oil, and Telcom. No dispute there.But I look at Level3 and go really, they are getting pricing at rates 20X more than they should? Where is the money going?
Therein lies the problem that plagues this industry; namely the lack of a framework that gets everyone on the same page analytically, conceptually and verbally. There is NO comparison between Level3 (principally layers 1-2 in the WAN and selectively in mid-mile WAN/MAN backhaul) and vertically integrated carriers like VZ and AT&T (wired and wireless). Zayo will likely do a reverse merger with LVLT at some point to drive scale economies to be competitive in local, regional and national bandwidth markets.Bandwidth is 20x-150x more expensive in the mid and last mile than it need be. For example a 100mbs backhaul circuit that costs $1400/month should cost $14 given current technology in the core and edge. When the FCC “deregulated” special access in 2002 they killed that market, and when they “deregulated” equal access in 2004 they killed last mile competition. When absolute bandwidth pricing only declines 5-10% per year versus 30-50% per year per bit, the result is the biggest arbitrage since 1984 (pre-divestiture) between retail and economic cost/bit.Google believes it can price 1 gbs at $70 because they have scale at many layers and boundary points and can drive marginal cost down uniformly. So there is hope there. And Steve Jobs resurrected equal access in the form of the smartphone.
Thanks for the reply, and good points, but how do you come up with $14/month? I understand pricing the equipment at the core and the edge, but I would think the greatest cost by far is the labor/permits/paperwork at the mid and last mile to connect the two. When I think of how long it took them to get the fiber to our building years ago, and how long it took them to wire the building with fiber, and how long it took them to then wire us, even though they are inefficient and even though they have high union costs $14/month would never allow a return on that investment.That is why I brought up Level3. It has been coming down drastically when you look at what you pay at the datacenter. The last mile has always been tough. We get VZ FIOS for business out here for $100/month for 100Mbs up/50Mbs down, we also get Comcast which is a tiny bit cheaper but not as much down, but they both are literally putting no more in. Either your building has it or it doesn’t. If it was wild-ass profitable why would they do that?. It might be for some non-economic reason like they want to hurt the unions that do the work or protect the video business.Not sure. It is an issue near and dear to my heart, I’ve always loved telecom and thought about how to start a business in it. As a matter of fact one of the co-founders of one of my businesses did, but it didn’t really work economically and he does solar now (which works only because of subsides)
Flat rate, all you can eat voice plans killed 3G roleouts to rural America. Data caps mean that urban transient usage (which effectively double intrinsic rural demand) can be monetized and provide for 4G buildouts in rural markets. Odd but true.
I’m surprised we don’t offer something similar for effectively “local” sites such as news, twitter, email. Metering is one thing, so is deep packet inspection, preventing basic use is another.
How does this outlook fit with mobile investments that aren’t necessarily open web? Android/iOS can (and do) provide improved experiences and therefore can enrich access to web information, but at the same time they are controlled by platform owners (to varying extents).
on a budget?book a bit pricey?YouTube to the rescue;http://www.youtube.com/watc…
Thank goodness (barely) for used books. Expensive, but not surprising.
The problem with this work and the opposing sides in the ITU debate is that we are comparing apples and oranges. On the one hand is the vertically integrated, government sustained monopoly communications model that is rife with inefficient subsidy and priced to reflect average costs. On the other is the horizontally layered, competitive service model born mostly out of large-scale private intranets where pricing reflects (rapidly declining) marginal costs.Without going into a long history of access charges (subsidies) and how the breakup of AT&T fostered what we know as the internet and how wired access is different (yet related) to wireless access, the solution is a balanced settlement system that fosters new service creation and is neither called party pays nor calling party pays exclusively.The balanced settlement solution needs to be end-to-end (horizontal or between networks) so that new entrants cannot be blocked. In the “IP” or “internet” storeand forward, mostly database look-up model, bill and keep sufficed. But in a real-time, two-way, bilateral application world we need reciprocal compensation.The settlement model also needs to work vertically or top to bottom (aka packet metering). There is a long history of the commercial user subsidizing retail user access (800, VPN, prepaid, ad sponsored internet and broadcast, etc…). Big data will ultimately be the “clearing mechanism” or currency that relates the commercial transaction to the communications event/session. The high-volume, low-marginal cost centralized procurer will demand accurate statistics on how much is being consumed and by whom.