The new SIPphone adapter has been launched. This shows how the vertical unbundling of telecom is accellerating. Traditional telecom attempts to bundle service (speak to someone not here) with access (copper cable to central office or local switch box). Arbitrageurs like Vonage are trying to bundle multiple bridging technologies (an ATA box to make your old phone work, and a PSTN interconnect). This product unbundles Vonage. I can now buy my customer equipment bridge (the SIPadapter) separately from my PSTN interconnect bridge (a calling card service which accepts inbound SIP connections).
However, I think that the SIPadapter people have yet to hit on the killer product. They need something that can disintermediate carriers without so much user effort. You shouldn't need two phones in your house. It should be trivial for the user provsion the device with the details of a calling card service, and have the device perform any pre-dialling or account details entry.
Another device might be necessary to help make this mass market. Many people will be slow to let go of their PSTN line. You should be able to plug in two uplinks: one to your cable/DSL line, and the other to the PSTN. The device would then automatically route over the right network. I dial my brother's SIP phone in Europe, it goes VoIP end-to-end; I call for a pizza, it goes through my local PSTN line; I dial my parents in London, it automatically uses a calling card prefix.
The days of Vonage are probably numbered because their own success will cause further arbitrage and disintermediation. I personally can't see myself paying their monthly recurring charge plus international calling charges for much longer: it is almost cheaper to buy all my relatives in Europe a DSL line and a SIP phone.
The Washington Post today discusses the vast tracts of cellular advertising on their largely undifferentiated products.
Here's the money quote:
"I find the cell phone industry incredibly annoying," said Henrichsen, a mortgage broker who moonlights as a pianist and carries a bare-bones cell phone. "They put way more options [on] than a sane person would need."
Henrichsen said he admires the ads but thinks the money could be better spent: "I'd rather that they work harder not to drop my call on the parkway."
Now, if we had an end-to-end IP approach, the network would have no concept that he was making a phone call, and calls would never be dropped when he walking into an elevator or drove through a tunnel. You would come back into coverage and just keep on talking.
Underneath the hood, technologies like CDMA 1xRTT and GPRS/EDGE are fundamentally tied to a circuit-switched architecture. Even when you have a packet connection, it's still really dial-up: channels have to be set up and torn down, and that creates massive (5-10 second) latency. You can't build a meaningful VoIP architecture on those systems.
The first carrier to launch an mobile network with true VoIP will clean up. To learn more about OFDM and its competitors just take a look at the sites of Flarion, Navini, or IPWireless. The time-out before ending a call due to loss of signal will become a user preference in the handset. The next step after "Can you hear me now?" is "The talk never stops."
Circuit-switched thinking runs deep. Because the only thing the PSTN can do is make a phone call, every problem is solved by making a phone call.
Take voicemail, for instance. Whilst trying not to bore you with the obvious, let's recap how traditional voicemail works. The phone network intercepts unanswered calls using a time-out, and re-directs them to a voicemail server. The caller talks to the voicemail server, and that server stores the message. The callee receives a message waiting indicator, and later on makes a phone call to the voicemail system. Everything's a phone call.
We're all so used to this paradigm, we never question whether any other mechanism is possible, or whether the current approach makes sense. We seem to assume that future technologies like multi-modal phones and CDMA EV-DV (simulaneous circuit voice and packet data) are needed before we will break out of this mess. That isn't necessarily so.
Assume a phone call requires an (extremely generous) 3Kb per second of audio. One hour of stored audio is about 10Mb of data. This is a pretty modest amount by the standards of modern flash memeory. Your mobile phone is perfectly capable of storing all your voicemail. The network is perfectly capable of transmitting the data in a sensible amount of time. Unlike email, most voicemail is listened to - the amount of wasted download is small.
Of course, the handset cannot be the voicemail server itself, because it may be out of signal range or turned off. Store-and-forward is still necessary.
You should be able to listen to voicemails on your plane journey home. You should be able to reply to them on a store-and-forward basis, even when you're not connected to the network. And most of all, you shouldn't have to use a clunky telephony user interface to navigate a message queue. And you shouldn't be restricted to one device for accessing your own data.
How often have you returned a call in voicemail, spoken to the caller, and then finished the call by closing your phone or pressing 'end'? You find yourself dumped out of the voicemail system (hey, you ended the call to the voicemail server - the phone couldn't tell the difference!). To go back and delete the message, you have to re-dial and navigate the whole voicemail user interface again. Yet there's a prefectly adequate 2 inch screen that could have been used to list the entries, just like an email inbox.
Or have you ever had the experience of someone calling you from their cell phone, leaving you a voicemail, but you can't quite pick out the critical number they dictated because their connection dropped out for a moment? But if their message had been recorded locally, and forwarded over a TCP connection, it would have been crystal clear.
Taking this further, why on earth is it the recipient's carrier that decides when and where to redirect me? Why can't my handset just look up in a directory where you want voice messages sent?
The technology to make the end points smart is here today. We might not be able to do reliable wireless VoIP on wide-area networks with the deployed technology of today. There is a usability gap still to be filled, and profit to be made from filling it.
The abandonment of the circuit-switched world for all communication that isn't both real-time and two-way is seriously overdue.
I've been reading The Innovator's Solution, and have a few thoughts on one of the side comments in the book (on p147-8). I'll write more on the book's central thesis and how it relates to telecom another time.
They examine the a priori approach of agreeing an industry standard (GSM) versus the market approach adopted in the USA. The argument put forward is that the reason wireless adoption in Europe is higher than the USA is little to do with this industry standard per se. Instead, the substitute produce (wireline telecom) is cheap in the USA, and the Europeans mandated caller party pays, making people more willing to hand out their phone number.
I would suggest there are deeper reasons to let the market decide. As a European living in the Midwest, you are sensitive to some major differences in the physical structure of the nation. The USA is big. Really big. When I flew home from Denver a few weeks ago across Kansas, you see endless scattered small communities. Not a big city in sight, despite crossing a distance in Europe that would have taken you within striking range of 5 or 6 conurbations. Population density is an order of magnitude lower than western Europe.
Buildings here are different. Most houses seem to be made from something barely more durable than cardboard. Everything is disposable (and when a tornado comes through town, everything is thoroughly disposed of). Notwithstanding the foreign perception of the USA as all being the Manhattan skyline and the Sears Tower, there are relatively few skyscrapers in most habitated parts of the USA. Land is cheap, everything sprawls into infinity.
The consequence is that the physical attributes of the radio system really matter. Coverage matters. Building penetration matters. Non line-of-site matters. GSM may be better or worse than CDMA, analog, etc. -- but that's for each market to decide. Rural areas are still often served by spectrum and power hungry analog cellular. That's not a problem, because the cows and corn fields aren't gagging to reuse that spectrum for anything else.
And for you end-to-end absolutists, remember that IP is just an abstraction. Vertical market integration of service and network happens for a reason. You need to be able to explain not just why a modular approach to the industry is beneficial, but also why buying connectivity separately from transport won't lose you some essential features that the IP abstraction hides away.
If this article on the FCC chairman Michael Powell is right, then the guy deserves a lot more support than he's currently getting. But if he really wants to protect the end-to-end principle of the Internet, there is a small but important thing he should do.
The words people use constrain the thoughts they can express. I would suggest that the use of the term Internet in marketing of access services be regulated. "Internet access" would imply a pure end-to-end pipe access. No ports or IP addresses blocked. Static IP address. No QoS meddling. No contractual limits on use of the service for hosting a home server. Anyone who tries to sell a corrupted pipe would have to use a regulated phrase like "restricted Internet access" or somesuch. Enforce a standard listing of those restrictions, just like credit card terms and conditions. Who would buy restricted access without first asking what those restrictions are.
Some leeway can be added in to enable networks to block ports etc. as a temporary means of protecting the network from transient attacks.
In some ways this reflects the Cluetrain concept of markets as conversations. We should not regulate what is said, but just regulate the semantics of certain phrases. Free milkshake with every burger doesn't leave any leeway in what I expect to receive. Internet access should be the same.
Let the market decide. But make sure that decision is fully informed.
Appeal to readership: help needed. What are the solutions to the following two problems, and how can a money-making opportunity be made out of solving them?
Instant messaging seems to be hitting a social barrier in the the affordances of the system. You're either my buddy, or you're not. I'm either chatting with you or I'm not. There's no middle ground.
It just often seems impossible to satisfactorily end a conversation with someone who hasn't left their keyboard. And I've got people on my buddy list who can still track my comings and goings, but I don't chat with them any more because our relationship was associated with circumstances that have passed (e.g. a work project last year). Yet, I don't want to drop them off the list for fear of causing offence and poisoning potential future interactions.
Where's the equvalent of edging away from the party bore in IM?
Via /. comes an interesting short article on the consequences of abundance and the erosion of economics based on scarcity. I'd challenge one small assertion:
The increased capability to communicate effortlessly with anyone in the world is an amazing result of technological know-how. It also means we can significantly reduce costs by sending work to the other side of the world, rather than employing people closer to home. Work is now geographically ambivalent and all white collar work is at risk of displacement.
At first this seems a seductively obvious truth, but I feel it is in fact a distortion. Yes, just as consumption has globalized (e.g. I might take vacations in Europe and eat European imported cheese), so has production. But much production is still highly correlated to geographical or cultural context.
As a product developer in the USA, I need to have empathy with the needs of the natives to be able to ideate good product solutions to the communications problems they face in their everyday lives. The marketing department needs to communicate to those customers and use social cues (sports teams, historical events, entertainment personalities) that they know will resonate with their audience.
If I want to create and sell pictures and paintings of the American landscape, I'm not going be very productive in Bangalore. And the demand for "local art", personalized to the tastes of its consumer is huge and unfulfilled.
Aesthetics are highly local, and if Virginia Postrel is right and her thesis holds true, then the service outsourcing mania will rapidly reach its limits.
The telecom spin? Don't assume that the death of distance means the end of geography. Your network and product ultimately deliver unique messages to unique people. And don't invest in too many fibers going into India until their own markets and local demand for imported data increases.
UPDATE: On cue, Om Malik reminds us that the world is analog and location still really matters. Incidentally, the former SVP of strategy at Sprint, Liane Pelletier, was recently, um, overwhelmed by the urge to seek new opportunities, and ended up in Alaska. Perhaps Alaska isn't the new telecom Siberia after all.
RING-RING. RING-RING.
3am. The phone goes off. RING-RING.
Uuuurrrrghhhhhhhhhhh. You pick it up. It goes dead. Again.
Now, this little scenario can take one of two paths. The first is the one you already know: next morning, you call the customer service department of your friendly local telco, and lodge a complaint. Their investigation bureau traces the call, aggregates data from multiple complainants, and initiates punishment to the nuisance caller.
The other path is the one you potentially face in future with an IP-based solution like Skype which doesn't have strong "reverse identity tracability" built in. Who do you complain to? Nobody. You can't easily recover the real identity of a perpetrator of a social nuisance on Skype. But that isn't real problem.
There is a reason Skype doesn't socially scale beyond an initial communitarian group of users. It is because no identity collateral is put up when initiating a communication on Skype. That collateral doesn't have to be traceable back to me personally. Even if I call you every night at 3am with a pre-paid cell phone that I bought for cash, a punishment can be exacted for abuse of the service: I can simply be cut off. I then lose the value of the phone, plus the number associated and the social connections it brings me. My caller ID wasn't spoofable or hidden.
The user identity of a PSTN telephony user is welded to the transport and its associated economic transactions (account or handset purchase): the exact opposite of the separation of connectivity and service that the end-to-end model suggests.
On Skype, Free World Dialup, and similar systems not linked to a biometric person or a reversible economic transaction (like provisioning a phone), you're stuck. In fact, Skype is worse than Free World Dialup - you aren't even being directly contacted by your correspondent (so no IP address to work from), and the protocol is completely proprietary so you can't decode what's going on.
In a sense, a phone number is a personal brand or identity that you place "at risk" by using it. The mere act of being issued a telephone number hooks you into a legal, regulatory and social framework.
The Internet is providing new means of creating "at risk" identity collateral. For example, this blog and the name "telepocalypse.net" has taken a certain amount of my time and money to produce. Abuse of the service (e.g. through libel, or copyright infringement) could result in a lawsuit and loss of that investment. It is possible to find out who the infringer is (little 'ol me!), even if all you have is their DNS record. If my blog name came up as my verified caller ID, it would still have the useful anti-abuse feature, if somewhat weakened.
Comment spam on blogs is another example of an open communication system that demands no identity collateral, and therefore has no recourse in the face of abuse.
Identity is central to making an open communication system socially scale. The E164 telephony numbering space has subtle value that differs from DNS or anonymous systems. The loss of that value might only be noticed after it has gone away and social ills start to afflict our communication ecosystem. Caveat communicator.
Following on from yesterday's discussion on how transport is a loss leader for getting people to engage in transactions, I thought I ought to point out the obvious regarding business models for public WiFi.
I believe that this market will fragment into three components:
These could also be thought of as being in increasing order of captivity. Relay of access through a persoanl DSL connection to the park outside appeals to a totally non-captive audience. Municipal wireless access will act likewise.
The fundamental nature of WiFi is its short-range nature, notwithstanding those with large Pringle can antenna collections. It is geo-centric. You have to go there to experience it. Companies pay large amounts of money to display commercial messages in public spaces (TV, billboards, magazines, etc.). In your own commercial domain, there is no charge to make a pitch at a potential customer. WiFi in this case isn't competing against other forms of access. It's economic model is in competiton from advertising.
When your audience is already captive, then you're going to have to adopt a different model. They have to be truly captive as this story shows. Then you can milk them to the point that outrage kicks in. People pay to go to the toilet at major UK railway stations, and they'll pay to access the Internet when they're already engaged in a transaction tens or hundreds of times as costly as the access charge.
Anyone interested in the telecom revolution should by now have heard of Skype. For those just returning from backpacking across Siberia, Skype is and end-to-end encrypted, fully distributed, proprietary, PC-based telephony system.
Now, like all other PC-based systems, it has some drawbacks. You turn your PC off, and while the awful din of hard drives and fans stops, so does your telephony service. Bad luck if you need to call an ambulance, too --- try yelling out the window instead. Turn the volume down on your speakers and you don't hear the phone ring. Plug your headset in and that turns your speakers off - bad luck if you're away from your PC and it rings. And so on.
One feature of Skype is the ability to enter extended personal information into its directory. This includes your preferred language. Our multi-national household includes my dear wife, a Lithuanian citizen. Now, you will be excused (by me, not my wife) if you're struggling to pull your altas off the shelf, because Lithuania is a green and pleasant (if slightly flat) country of just 3.5m people, the great majority of whom speak Lithuanian. A modest Lithuanian diaspora is spread thoughout the world, many concentrated in a few northern US cities.
What does my strange family situation have to do with telecom business strategy? Well, we've been leaving Skype running for several weeks now. We only use it in anger to talk to my brother in London (I am in Kansas). But my wife has received a half dozen unsolicited calls -- and all from people listing their language as Lithuanian. (I've only received one such call, from a 48 year old Swedish woman. I didn't answer...) The human urge to connect and communicate is clearly strong.
You could argue that Skype isn't even competing against the PSTN system. It is really, for narrow sub-pupulations, a degenerate version of Friendster and other social networks.
The lesson of this is that the potential value proposition is much, much broader than dial-a-number and talk. The obsolescence of the PSTN will be caused not by cheap traditional voice run over the Internet, but by services the PSTN cannot ever deliver.
Airlines run networks, too. Whilst the delivery of 80kg packets of blubbering human flesh may not be directly comparable to telco networks, there are some great similarities as well as important lessons. Airlines are becoming more deregulated than telcos, with lower barriers to entry to creating a new network (since planes are notoriously portable in a way that fiber and copper aren't).
Firstly, consider the case of BAA, which runs most major UK airports, including Heathrow. Of BAA's £1.7bn UK revenue last year, £743m was from retailing and other ancilliary service; only £665m came from charging for flying people around.
Ryanair are trying an even better trick, of extracting subsidies from regional airports merely for delivering passengers to local businesses. Their only problem seems to be European rules which prevent such overtly discriminatory subsidies. They are heading towards a model of "reverse landing fees", where they get paid to deliver wallets within reach of geographically-bound businesses.
The Scottish Executive has been busy spending taxpayer pounds on airline subsidies to encourage new routes to open up.
At last, Glasgow lands Europe flights bonanza ... Mr Dalrymple still hopes to secure support from the fund for the new routes and stresses that existing flights from Italy, France and Spain have brought significant numbers of tourists into Scotland.
For example, Continental are being given a bung just to fly between Edinburgh and New York.
The patterns is that stand-alone transport is becoming a loss leader to generate captive audiences for other forms of commerce. Telcos need to work out who their customers are, why they are communicating, and insert themselves into the value chain through means other that coercion and regulation. Failure to do this will result in extinction.
I've spotted a few small instances of telcos doing some minor innovation in pricing and bundling models for otherwise undifferentiated access businesses. This could become a pattern. First, The Register has a report on how BT are doing tiered access charges for WiFi. T-Mobile in the UK have their MixIt plans (Flash beware!) which mix pre- and post-paid. And of course, Verizon in New York is bundling, ahem, "free" WiFi with regular DSL access.
Could we be about to see an explosion of new pricing plans for metered access which mix time, geography, access technology, and payment methods?
Andre Durand and others have been creating a classification of identity. Roger Clarke has been working on models for privacy. Brian Blaser has a breakdown of types of reputation (i.e. trusted identity). These are useful classification systems. But I feel there is a need to have another axis on our map, that links the type of identity data to its potential economic value. Any organization with ambitions to become an identity broker utility (think: struggling telco) can then do an inventory on its identity data and see where the unrealized value may lie in those customer data assets.
The model I am proposing has some similarities to the "mydentity" model which runs on the X-axis. That model (in my favorite form) has increasing levels of "personalization", to recycle and over-abused buzzword. These scale through the following:
Some people combine layers 0 and 1 (which loses an important distinction, in my opinion). Some think that layer three should be called the "marketing identity", but that seems too narrow and prejudicial a definition to me.
My value classification extends this to include a Y-axis, with increasing levels of certitude:
What do these mean? Read on.
Level 0
The base level is "untraceable in space and time". That means it cannot be traced to me, and I can supply different data at different times without the fact being detected. So the fact that I am 32 and male can be revealed to any 3rd party, and as long as they can't see me in person or hear me then I can equally claim to be a 23 year old woman tomorrow. There are no consequences of lying repeatedly.
Level 1
Level one data is as anonymous as level zero, but is traceable over time. Essentially, you have acquired a pseudonym. Cookies are the classic example of this. A web site like Priceline that wants to prevent you from submitting multiple identical bids (themselves level 0 data) uses a cookie to track you. You can lie, just not over and over.
Level 2
At the next level the data is at least traceable back to someone. You can lie, but it has consequences. You might also not be lying, but could unintentionally give inaccurate data. You can give your name, employee number, previous address. Someone may act on this data, for example by cutting a pay check or shipping a book to you.
Level 3
We add in validation to the data, also known as data quality. This is subtly different from level 2, and moving identity data from level 2 to level 3 is already big business. If I store my data in a profile, and keep giving it out, then it is more likely to be correct than if I have to enter it fresh each time. If someone checks my ZIP code against an address database, the data may not change one iota, but the meta data now starts to take on value.
Level 4
This is the ultimate level, where we add in verification. A trusted third party asserts that the data is true to some degree, in the context of some liability relationship for that assertion. The American Express logo on my credit card acts as an assurance to a merchant that they will be paid in return for entering my credit card number into their system. The assertion is relatively weak, since traditional magnetic stripe cards are easily forged. The credit card number has a check digit and is printed on a card, which at least makes for strong level 3 data.
It is interesting to observe that "theirdentity" is incompatible with a value beyond level 1. At best they can know a product was pitched to you last week. This value model helps us understand why mass market advertising is in such trouble. It also shows up the weakness in Google's identity model: they are stuck at level 0 (what do you want to search for today?), and haven't yet made it to level 1 (what should I return to you based on what you searched for before)? A next-generation search engine might look to exploit stronger identity data about both the searcher and the data sources.
So, what's the link to the future of the telecommunications industry? Ah, I'll keep that for the next installment...
Wandering through the office buildings of the telco I work for, I happened to notice a ceiling panel had been removed for maintenance access. Slinking its hidden way through the heavens was a neatly tied up bundle of -- at a guess -- at least 30 Ethernet cables, snaking their way off to the cube farm. This set me thinking: is the local loop mentaility running much deeper than previously thought?
Creating a physical link for every cube creates problems. The layout of the offices becomes extremely inflexible. Add a new device and someone has to crawl up in the roof to route a new cable through. Why isn't there a simple switch associated with every cluser of cubes? (Available from your nearest Best Buy for a few tens of dollars - cheaper than a large bunch of Ethernet cables run for tens of meters.)
Of course, allowing a flexible office layout where the local inhabitants can defy the cube police is almost certainly not corporate policy.
This same problem of local-loopism, of course, afflicts DSL with distance limits from the central office. The Internet is 30+ years old as a concept. Will the Internet's distributed network technologies finally push aside legacy thinking before it gets to collect its pension and free bus pass?
I was thinking during one talk today on federated identity how there could be a need for a whole new class of infomediaries to make identity federation scale. If you're Wal-Mart and you have 30,000 suppliers, you aren't going to want to manage access to your supply chain extranet applications yourself.
Even if the remote authentication assertion is done at the role level rather than named individuals ("someone allowed to access the monthly sales figures was authenticated"), just managing the PKI infrastructure and general systems management is non-trivial. Perhaps some systems integrators will step into the breach, or maybe this is an opportunity for new companies to form.
I wonder what buzzwords will be coined for identity value chain integration specialists?
One session from Digital ID World was on trusted computing. Cory Doctorow gave an impassioned and rapid-fire defence of the balance between fair use and content creator rights. The core was an attack on remote attestation, and in particular a recap of the recent EFF paper on the provision of an "owner over-ride" to allow fake attestations. The clearly spoken (if somewhat stiff-mannered) Peter Biddle from Microsoft avoided head-on engagement on the issue.
To me, both missed the biggest issue, and the one that would really gather political attention in a way that fair use doesn't. If NGSCB takes off, some time around 2008 we can look forward to MSBlast.NGSCB, which will take advantage of some "trusted" (but flawed) code, and generally causes mayhem by reproducing fast (totally trusted, remember!) and will be hard to squash.
Remote attestation is, in the awkward new terminology, a homeland security problem.
I'm currently at the Digital ID World conference, where a number of the Internet illuminati are also blogging away. The line-up so far has been impressive.
The thing that has struck me most so far is the near-total disconnect between identity in the context of enterprises, and that of what you might call "public identity". (I'm desperately veering around to avoid the word 'consumer'). The technology problems of making verifiable assertions backed up by authorities do not map one-for-one to the human problems of trust.
Within the enterprise environment, the trust relationship is well-defined, with certain job functions and roles being granted access to business entities within the corporation.
Employees have a defined relationship with their employer. A strong hierarchy provides directionality to the trust vectors. Of course, recent history has shown that a Cxx job title is no guarantee of probity, but the model is a reasonable approximation.
As was pointed out in the sessions, identity is only of value in a particular trust context. You might not trust your stock broker to act as your baby sitter. Or the old favorite, you trust someone not to cheat at cards, but you don't trust them not to cheat on their spouse. People have extremely complex and dynamic relationships. This contrasts with employees or other narrow role-based users, where the context is singular and relatively static.
The consequence of this is that even the very language we use to talk about identity falls apart when you move out of the enterprise, because it muddies and confuses the richness of the real world. For example, 'trusted computing' makes sense in the context of a corporation. The user's laptop is owned by the corporation, and there is no tension between the wishes of the device's owner and the policy creator. The laptop is assigned to a single employee. The computer and the user are almost interchangeable. "Trusted computing" equals "trusted users".
This isn't so for you and me. With technology like remote attestation, the trust relationship isn't just between me and my PC. It is painted as being between my PC and other computers. My PC is supposed to talk to my laptop and they trust each other. But this is ludicrous! My PC has no concept of trust -- or at least, until the day I come home from work and my PC and laptop have a marriage certificate proving their undying love for each other, I'm going to remain a skeptic. Only people can create and maintain trust. Computers are neither trusted nor trusting, merely authenticated and authorized. "Trusted computing" in the public context means "distrust strangers".
Trust is projected within enterprises by the use of policies. But in the world of people, trust decisions are often made on the basis of brand. A corporation will trust Internet Explorer for a secure transaction because the target site has a digital certificate with a root issuer like VeriSign. The average Windows user has never heard of VeriSign, and probably doesn't know what a digital certificate is. At best, they trust their banking session to be secure because Microsoft painted a padlock icon, and they trust Microsoft and their bank to work together.
The telco business model perspective is not hard to fathom. Telcos are major identity businesses. They issue tens of millions of identifiers (phone numbers), millions of physical devices with unique IDs (cell phones) and operate dozens of applications. The relationships between these are all juggled in some vast provisioning and customer data system.
Yet to date, there has been a singular failure to build external-facing identity businesses. These would, under user control, allow the brokering of user data (name, address) and meta-data (is this Martin's real address?). Telcos, when they make money, do it by offering large-scale horizontal communications applications. If public identity services aren't such a thing, I don't know what is.
Imagine a future where you can phone up Best Buy and ask to buy your shiny new plasma TV. You don't have to dictate your name, address or even credit card details. You either have to press a few buttons or enter a PIN to authorize release of this information from your telco to the shop. Even better, the telco validates to the store that this is your real address (we installed the phone line) and your real name (we saw your driving license when you opened the account) and your credit card (we used it to bill you every month). So the store not only saves time at the call center, but the address is correct (no mis-deliveries) and validated (no fraud). Maybe the store doesn't even get given the credit card number, just a token proving that the card was debited. No more propagation of sensitive data, no more liability on the store to securely hold it. The 10 cent phone call turns into the 10 dollar identity brokerage event.
To get there the whole identity conversation needs to move on a level beyond identity federation (which doesn't scale well as a business, because it requires too many bi-lateral agreements). Instead we need an identity projection model, where the user's data is truly shared under the user's control.
I recently borrowed Strategy Pure and Simple II by Michel Robert from a work colleague. Although the book crams 50 pages of content into 300 pages of paper, it does have some useful insights.
In particular, it lists archetypes for corporate strategies. The hypothesis is that you can only be following one of the following ten core paths (with examples):
For example, a product-driven company will try to find people -- of any class or type -- whose needs are filled by the product. On the other hand, a user-class driven company will source whatever products are needed to keep their clearly defined customer base happy. If the customer needs change, so will the product portfolio.
You will note that traditional telephony falls into the unique distribution method category. The only way of obtaining the "speak to someone not here" application was to partake in the bundled distribution network. Clearly, that is becoming no longer the case. This creates a potentially irresistible force tearing apart the vertically integrated telcos.
At the bottom of the stack is access or connectivity. This is driven by a capacity-led strategy. Every bit-moment of fiber and spectrum needs to be used, else it is wasted forever. In fact, it is subtly more complex than this. When the network is bought as a unified whole on a subscription basis with a captive audience, the driver is to minimize usage (since usage drives cost, and revenues are fixed). On the other hand, access can be metered, or provided through some aggregated provisioning system (like Wayport or Boingo, of even traditional cellular with affiliates and roaming). In this case usage drives revenue, and hence maximum capacity utilization is the key.
Above this in the stack is the application platform. In traditional telephony this is all the SS7 switching, voicemail storage, etc. This is probably a technology-led business, although you could argue that proprietary platforms like Windows are growth-led business. (The only secret in the secret sauce is that you've got more of it.)
Finally, there is the application and service layer. This could be driven by a number of different strategies, although product is by far the most likely.
Hence there are really at least three businesses struggling to escape from the traditional telco. Few if any telcos are organized as business units around these areas. The result is likely to be under-utilization of network resources and excessive build-out and capacity. Failure to re-structure will perpetuate the over-capacity of the industry because the network owners are not strongly incentivized to find new network uses, since the new applications may compete with existing revenue streams. The differing strategic needs of the components of the business are likely to compromise any of them from being effective.
The problem is also likely to get worse as technology improves. For analog cellular, the whole system was totally tailored to the needs of voice telephony, and packet data was out of the question. It wasn't really possible to sell access unbundled from the telephony platform and service.
With 2.5G and 3G networks, some limited real-time packet applications are possible, but the network is still strongly vertically integrated with the voice application.
4G-type technologies (the OFDM family) will enable a totally IP-based architecture. This may well be the straw that breaks the camel's back.
In conclusion, instead of seeing industry consolidation, expect to see delamination. There isn't a good reason to build half a dozen nationwide IP network which are then also in competition with metropolital and local wireless networks. A few wholesale networks are likely to underpin many Mobile Virtual Network Operators (MVNOs). Other parts of the value chain may also be re-organized. A product-led MVNO might outsource customer care, while a user-category driven company will have a nugatory product development department.
Many of the Cluetrained will already be familiar with the arguments about the Internet being a peer-based communication system, not a content broadcasting medium.
One great little data point I picked up from this fascinating (if long) interview on municipal broadband:
In fact, in ultra-rural Grant County, WA, where users of the County's FTTH system have affordable access to speeds of 100 Mbps in both directions, bandwidth usage has jumped more than 600 percent and upstream usage actually exceeds downstream usage. Why? The County believes that small businesses are sending substantially more information to the Internet than they are downloading, and gamers are vastly increasing their real-time usage.
So, it really does seem like asymmetric network bandwidth is doomed. Why can't we have systems that allocate you a fair proportion of the municipal pipe, and leave the balance of upstream vs. downstream to the user? On an Ethernet-style system, why does it make the slightest bit of difference whether the packet is being sent or received?
Few can have failed to notice the hubbub over Verisign's SiteFinder "service", which returns a DNS record pointing to their pay-per-click search engine for every unresolved DNS lookup.
There have been many good write-ups on the technical reasons why this is a bad idea. It breaks a number of existing systems.
More fundamentally, it violates the end-to-end principle. The end points should have the ability to determine the appropriate action associated with a lookup of a non-existent domain. This is alluded to in the article at CircleID:
VeriSign justifies SiteFinder by calling it a "service" and says it helps users by giving them a friendly search page instead of a non-existent domain message. Some modern Internet browsers such as Internet Explorer see this non-existent error message and instead return their own similar search page. If an Internet Explorer user doesn't want to see Microsoft's default search page they could install different plug-ins to handle the DNS error, or the user could choose a different browser. But in either case, the user has an option.
Perhaps ICANN need to consider a clause in their next contract with (hopefully) the replacement for VeriSign stating that the TLD operator must do nothing that violates or damages the end-to-end nature of the Internet?
Whilst re-reading an article by Bob Frankson on QoS, I was thinking about the difference between what users value from the telephony application (speak to someone not physically present) and what the telcos value (connect devices through a network with 99.999% availability).
The uptime of the application (as opposed to the network) is the chance you can communicate with your intended target given that the target is physically willing and able to receive a phone call. But if they have the TV turned up loud and don't hear the phone ringing then the network might as well not be there. If they're out of the house and you phone their landline instead of their cellphone, you get 0% availability. In many ways, a small amount of presence or location information can make a big difference to the "true" application availability. The cost-effectiveness of the last few 9's of network uptime is questionable. The system is only as good as its weakest link.
The end-to-end perspective is that you have to view the system from the perspective of the user, and not any intermediary. The value of the system is defined by the users at the end points.
There's a great little report over at Vonage's PR department from UBS's research division. Here's some of the choice quotes:
We believe VOIP technology has the potential to do to wireline carriers what file sharing is doing to the recording industry.
Perhaps I should get that printed onto a T-shirt?
In Japan, VOIP-based competition has emerged, allowing consumers to save as much as 90% off of usage. Calls to other VOIP customers are indeed free. As a result, VOIP-based telephony competitors now provide service to more than 10% of households and 45% of high-end, broadband households after less than 2 years of service while incumbent NTT is feeling the pressure.
Or how about this:
In FY2002, NTT East and West saw minutes decline 29% for total analog and ISDN traffic, up from the already significant 21% decline experienced in FY2001. The total number of lines at NTT East and West has also been slowly falling by 1.6% in FY2001 and 1.7% in FY2002. We estimate that losing fixed-line long-distance and international calls to VOIP, could hurt 7% of NTT's consolidated revenues, or about 13% of fixed line revenues.
Unfortunately, UBS have totally missed the point of the Internet: it isn't about cheap telephony, but about improved telephony (and improved everything else). For example, should US citizens get tired of John Ashcroft and friends eavesdropping on every telephone call, or there is a single scandal from unauthorized data gathering and wiretapping, then the demand for encrypted telephony could skyrocket. The PSTN can't deliver this.
They need to read the business classic from Clayton Christensen, The Innovator's Dilemma. This would show them that the real reason the Bells and other telcos will fail is because the market moves on to value new things (encryption, service integration, new codecs, etc.) while the incumbents exceed the market needs for what they believe the customer values (ever cheaper plain old telephony).
The business model for most current Internet telephony companies is flawed from the outset.
Will Vonage, the leader of the pack, have a long-term future as a business? Well, that depends on them executing a strategic shift as they move out of their growth phase (still some time away). They may have already blown it.
At home, we use Vonage as our landline telephony service. For those who don't follow the world of IP-based telephony providers, Vonage is a company that gives you the equivalent of full-service landline telephony, just using an Internet connection and an telephone adaptor box. I'll write up our experiences some other day, but for now all you need to know is that Vonage is the clear market leader in broadband telephony in the USA.
What's the problem with their business model? Two things.
Revenge of the Regulators
Firstly, they were trying to arbitrage the regulatory environment. There are major costs associated with obtaining a traditional land line or long distance service. These include emergency service fees, universal service taxes, various random carrier cost-recovery fees assocaiated with regulatory impositions like number portability, and local and state taxes.
Vonage are currently experiencing a blizzard of regulatory assaults in establishing their PSTN look-alike service throughout the USA.
Since their VoIP service looks like the PSTN and smells like the PSTN they are having a hard time arguing that the existing regulations should apply to the technology (analog copper lines) rather than the application (speak to someone not present addressed by a telephone number). This regulatory cost advantage may be unsustainable, and regulatory fees are already appearing on Vonage bills.
A ceiling on their potential success
More fundamentally, their business model is about arbitrage of the access fee component of traditional telephony. Because there was only a single pipe into each household, and that pipe had a single purpose (PSTN telephony), monopoly rents could be extracted. But the margin of advantage that Vonage offers will be squeezed by the success of their own business model: the more people who join Vonage, the greater the pricing pressure there will be on incumbents. Furthermore, Vonage will have to build out an infrastructure in proportion to the number of subscribers they receive. The incumbents already have their equipment and business support services in place and paid for: they only need cut back on operational costs.
Vonage should re-define the market
The shift Vonage need to make is to stop trying to copy the PSTN on IP. The whole point of the Internet is to enable unforseen innovation at the end points. Instead, they need to take a leaf out of Skype's book and do something that the PSTN cannot do. This could be integration with presence, encryption, higher voice quality, voicemail stored at the end points, etc.
They may find this switch hard or impossible to execute on because the Cisco ATA-186 adapter boxes are so inflexible and limited in their functionality.
Ultimately, the future of telephony belongs to a centralized directory service provider such as MSN, Yahoo! or AOL, or a decentralized system such as Skype. Vonage will get their footnote in the history books -- and maybe even a chapter -- but they won't be the victor writing the history of the new telephony.