Wednesday, May 25, 2005

More bread for the ducks

Martin Geddes - Telepocalypse

UPDATE: Warning. This is a long rambling article and I come dangerously close to totally change my mind at the end. Several times.
UPDATE: And before you embark on the long, dull journey through my turgid prose on public service calling, remember that the only real solution is to put some onus back on the connectivity provider, because anything above the IP abstraction layer can’t — by definition — locate itself reliably as long as the devices lack reliable internal location technology. Geography only exists at layer 0, so whoever does layer 1 is in the position to pass it on.
The debate rolls on…
A big problem with regulating 911 service is deciding who you are trying to protect. Is it the purchaser of the service, or the user of the emergency service? They are not the same, and this can lead to different answers.
If it is the user, then you have to focus on the external “implied semantics” of the device because that’s all the babysitter has got to look at when little Johnny stops breathing and goes blue. She isn’t going to boot your PC to see what telephony apps you have installed. At the very least she mustn’t waste time attempting a VoIP 911 call when she should be rushing out to the neighbours or using her cell phone instead.
Now, whether homes should be mandated to provide 911 service when strangers are allowed in unsupervised is a different matter; it’s similar to enforcing public liability insurance cover for corporations. If you took the “protect the public” argument to its logical conclusion, that’s the result. Yet it is probably impractical and unenforceable.
If someone insists on putting a POTS look-alike SIP phone attached to FWD in their living room, and ‘911’ is mapped as a short code to call Uncle Fred in Montana, and the babysitter gets confused, that’s just tough. Before we installed Vonage we didn’t have any landline-like phone in our apartment in Kansas City, and if we took both our cell phones with us, there was zero phone service beyond the PC and Net2Phone. (Then again, we didn’t have a baby before we got Vonage, so I’d have been very alarmed to find a babysitter in my home.) Forcing people to get a phone line seems a bit rich; better to have the babysitter check the place out first before accepting the work.
You cannot rely on anything in the contract or at the point of sale for protecting third parties. [The babysitter example is slightly flawed because the baby is associated with the first party, and the third party babysitter isn’t the one in distress. No matter, the principle is the same.] We’re dealing with a market failure here; the in extremis nature of emergency calling means the market needs a minimum level of mandated functionality.
On the other hand, you could say you’re trying to protect the buyer. This is very different, because then the whole consumer protection at the point of sale or activation comes into play. You could mandate that all PSTN-interconnected VoIP services simply play a “this system does not offer emergency calling” message prior to allowing the first call; the vocal equivalent of Skype’s pop-up. This would answer the “well, what did you expect” problem of most of the calamities currently being bandied about.
Although consumer protection at the point of sale is clearly important, it doesn’t lend itself to mandating 911 service. What if you just wanted a cheap second line without the expense of 911 service? The free market libertarian in me feels queasy at the though of such forced bundling. But E911 also forms an essential public good; you might call because you can see my house being burgled. I can also think of many examples where the buyer and user are not the same. Just think of shared households such as students, or landlords vs tenants, or temporary accommodation. So forced tying of 911 to PSTN calling isn’t unreasonable, even if reasonable people could disagree.
Interestingly, in the above referenced letter from Skype’s attorneys to the FCC they include both the marketing of the product as well as the existence of consumer premises equipment with dial tone. To me, you can only do the duck test by looking at the supposed telephone duck, and not the marketing river.
So, onwards to some concete examples…
Free World Dialup, as Aswath notes, doesn’t use PSTN numbering/interconnect, so no issue.
Let’s refine the duck test definition a bit. Perfection is the enemy of the good, and today we’ve got real bad in terms of the gap between public expectation and reality of what they’re sold. Smallprint won’t get you out of the semantic mess of expectations that words like “phone” induce. (Think: “The Broadband Phone Company” - ie Vonage - vs Skype which makes no tagline promises about telephony.) If someone sold you a “car” where some getout clause in the contract said “not suitable for use on public highways”, you would expect to sue for false advertising.
If you can plug in a POTS phone into the “out of the box” configuration, and it uses PSTN interconnect, game over. E911 is mandatory. That covers 90%+ of the relevant market for the next two years. No definitional issues around the service or its marketing.
Next come devices which look like POTS phones but have extra smarts to enable them to make PSTN calls via non-PSTN means. Any PSTN interconnected service (outbound calling) requiring a user-owned input device whose sole or main function is telephony, offering a 0-9*# keypad (or equivalent numeric entry system), and providing dialtone, that’s a duck. In other words, a different plug won’t save you.
Now we get into a messier area where the bits are unbundled into different boxes. If the microphone and speaker come as accessories connected via dedicated inputs, it’s a duck.
Skype in an ATA with SkypeOut is a duck, because you can plug a POTS phone in. (Do you need SkypeIn to offer call-back? No, the babysitter can’t tell. But the PSAP needs to know there’s no call-back and keep the caller connected.) Skype in a router isn’t a duck, because it can be called via a PC. If the user conspires to add a string of adapters to use a POTS phone, that’s their bad luck. Use of a POTS-alike phone has to be a requirement, not merely a possibility, for such a severe mandate to apply.
There’s no need to define “PC”, because you only define what is subject to E911. No need to regulate tin cans and string with a printed warning label “Danger: Not Suitable For Emergency Calling.”
If you’re in a grey area, and the court catches you out, tough luck. You needed to put clear blue water between yourself and the PSTN to avoid the PSTN regulatory tar pit. Life and death is too important.
UPDATE: Some more examples. What if company A sells a POTS phone, B sells an ATA, C sells a SIP-to-PSTN gateway service (little different from SkypeOut)? Is it a duck? Can we get away from how it is marketed — in aggregate or individually? Ultimately you can’t prevent people putting devices that look like PSTN phones in their home but don’t do 911 calling. You can’t make them have devices that ring, or have the volume unmuted, just so they can receive callbacks. It could just be a toy phone! The protection of third parties is always going to be incomplete. In this case I say it’s not a duck. You can only work on enforcing consumer protection at the point of sale and/or activation of such a disaggregated product. There will be an increasing proliferation of ways of reaching the PSTN. It’s impossible to control all of them.
What if Vonage enable you to make calls via their web site as well as via the ATA? Doesn’t this dilute the “requirement” bit above? I should probably refine the definition to something like “offers at least one physical port/interface that requires a POTS-alike phone”.
Maybe another difference is in that SIP gateway vs. Skype’s closed protocol. Skype can control what gets interfaced, and can argue you can never acquire a device that could cause babysiter confusion. (I’ve not looked at the Skype-embedded products in detail, so can’t say whether this currently holds true.) Thus they should not be mandated to include any form of 911, in addition to the practical reasons they list. The “open” SIP gateway that enables (but does not require) a POTS-alike device is one step closer to regulation.
Ultimately we have to go back to first principles and re-engineer the whole thing from the ground up. But it’ll have to become a lot more broken first before the political pressure forces anything to happen. I wonder if private-sector response services will emerge to take advantage of the increasing gap between the possible and the actual? Emergency service calling is truly pathetic compared to what we can do or need. If I’d had an accident when I was in Italy, I don’t need the nearest help point; I need the nearest one that has someone who speaks English. My cell phone’s interface is set to “English”. The potential is there.
Or is the co-ordination problem too large and the benefit too indirect and diffuse for a market soltion at all? This one will run and run.
UPDATE: You have to take a lot of this issue with a large anti-competitive dose of salt. Cell phones have got away for years with weak 911 serivce. Cordless phone makers don’t have to shout about how your phone won’t work when the power’s out; they aren’t all forced to work off rechargeable batteries. Telcos don’t want to see 911 service suddenty start advancing by leaps and bounds on the stupid network, as it could, ahem, set a rather bad precedent. Regulatory spending massively outstrips R&D spending; after all, it’s your core competence.


Post a Comment

<< Home