Misunderstanding natural monopoly

Seth Godin’s blog is a great source of pithy, wise, generous insights. I read pretty much every entry, and often find myself going “Huh! I’d never thought of it that way”. But as usual. I’m only going to blog about him when I disagree.


In a piece written a fortnight ago, Godin writes:

Why is there only one Twitter? One centralized phone network?

A natural monopoly is a business that benefits its users by being the one and only. If there were two incompatible phone networks, you’d need access to both in order to call the people in your life–and remember who was on each network.

That’s true as far as it goes. It certainly explains why there is only one eBay. But if it was a fully general explanation, the inevitable conclusion would be that there would be only one email provider. Thankfully it ain’t so.

Email wins — it’s survived everything that’s been thrown at it for thirty years — because it’s embodied not as a specific service but as a protocol that anyone can implement. I can run my own SMTP server if I want. Heck, I can write my own SMTP server, the specification is freely available. Because of that, no one company can impose a stranglehold. No-one can decide one day that all emails will carry advertisements. If any given email provider does decide that, its users can just move away.

Not so with Twitter, which is why I am ambivalent about it despite its astonishing ability to connect people. (Well, that’s one reason. The other is its astonishing ability to absorb time.) If you want to tweet, and read others’ tweets, then the only place to do that is Twitter (or at best a different UI built on their API). They control the service. And that can’t be good in the long term.

EPSON DSC picture

So: is it possible to build a decentralised Twitter-alike, based on a protocol rather than a service? I guess it must be, since tweets are just tiny blog-posts, and the blogging network is nicely decentralised. Pingbacks work between blogs hosted by different providers, for example. And you can make a unified feed of blogs by using their RSS feeds.

In fact, isn’t the infrastructure already there? Isn’t it just RSS? I can read all my friends decentralised tweet-like messages just by watching an RSS feed that merges all their feeds; and I can post my own tweet-like messages as tiny blog-posts which they in turn can see in their own merged RSS feeds.

All it takes is some simple front-end tools for reading and posting. We should make those tools.



13 responses to “Misunderstanding natural monopoly

  1. Just yesterday I was reading about a kickstarter campaign that was going to do exactly what you’ve described here http://www.kickstarter.com/projects/1904431672/trsst-a-distributed-secure-blog-platform-for-the-o. I haven’t backed it personally.

  2. Well maybe I’m just misunderstanding Godin, but (semi-)seriously. His point seems to me to be so flawed as to make me question whether he’s being paid by the Koch brothers.

    Ok, so I’m a raging liberal. But… “A natural monopoly is a business that benefits its users by being the one and only. ” ???

    Yeah. Just like how Acme Corp is the monopoly that issues all the TCP/IP protocols.

    I understand the strong need for agreed upon standards. And I’m aware ICANN has a monopoly on the domain names.

    It’s just the paragraph you cited makes it sound as if natural monopolies are in everyone’s best interests.

    That’s dumb. Monopolies, natural or otherwise, are in the best interests of the monopolistic organization in question. (duh)

    What would be in the public’s best interests (maybe, for the case of phones for examples) would be multiple companies that have agreed or been mandated to use compatible technologies.

    Of course there is the dread specter of the problem of no innovation and stifling of the free market. Sometimes, that is a real concern. Sometimes it’s a bit of a red herring.

    But don’t anybody come to me and my liberal biases trying to tell me that natural monopolies are “best” or even “good”. “Necessary evil” sometimes maybe, but never “good”.

    Furry cows moo and decompress.

  3. How would a decentralized twitter be fundamentally different from NNTP? Restrict each group (e.g., soc.twitter.neiltyson) to only one author, limit posts to an arbitrary length, and prohibit replying.

  4. Xavier Blondel


    The idea sounds quite interesting, but two points disturb me.

    First, using a decentralized system such as RSS or NNTP would complicate all of the search features (think about hashtags); you would need to either store everything on one computer (or a farm) to perform the search, or use a spanning search across multiple distributed sites. Definitely not impossible, but a hell of a giant interesting problem to tackle.

    Second, you would need to grow a substantial user base for the whole project to really take off (the usual chicken-and-egg problem: why should one join a social network where no obe is). This can possibly be overcome by focusing, at first, on a specific population, by solving some of their specific needs (something along the lines of a twitter for, say, biologists, with
    special features to deal with sauropod vertebrae – whatever they may be), then grow your user base from here.

  5. You might want to look at the Tent protocol, which aims to be a protocol for microblogging.

  6. Lots of interesting observations here — thanks to all who have commented.

    Seems that more has been done in this area already than I’d realised. It’s great that Trsst is headed towards funding (and looks like it may well make it), and that StatusNet exists, and that there’s Tent. No doubt these are technically superior solutions to just using RSS — or NNTP, as Colin suggests. But I’m inclined to think it’ll be easier to get this off the ground using an existing and widely-deployed technology than with a new one. I admire the security-consciousness of Trsst, but they seem a bit misplaced in a protocol that’s all about broadcasting.

    Xavier is right that decentralised searching is going to be a problem. I admit it’s an aspect I’d just not thought about. I guess I should read about what the Trsst, StatusNet and Tent specifications have to say about that.

    As for getting it to take off — in the way that Diaspora is so far failing to do — of course, that is a real social engineering problem. I like Xavier’s approach of starting with a niche. One important thing to do is make a gateway between the new thing and Twitter itself, so that people posting messages on the new thing will be visible on the current monolithic system.

  7. Let me take a step back here.

    The need of people is not to read tweets, but to communicate. Twitter has come with a system that, due its implementation (140 characters, easy to discover people, easy to distribute tweets, asymmetric following) is very useful for some things.

    Most definitely, Twitter is not a monopoly. You can communicate without Twitter (emails, blogs, IM, Google+, etc). Sure Twitter is convenient, but it’s not the only way of accessing the service.
    Ebay is, again, not a monopoly. You can still buy and sell stuff online on Craiglist or any other online shop. Again, is convenient, but is not exclusive.
    Phone networks are also NOT a monopoly. The networks are compatible and there is only one way of assigning the numbers. But I can choose among different providers.
    The concept of monopoly needs the lack of options to get a service. E.g. in rural areas, the choice of electric company is usually not present. If you want electricity, there’s only one company supplying it on your zone. We can endlessly debate if there are ways of replacing the service (using generators), if it’s enforced or just the high entry barriers makes it too expensive for other companies… But we can agree that you cannot easily choose. That’s the key point for monopolies.
    In Twitter case, I think that’s not the case. You can create an account at G+ and move all your activity there. Network effects are important in Internet, but they are not something set in stone. People do move for one service to another.

  8. The typical natural monopoly is a network. As Godin noted, it only makes sense to have one set of roads, pipes or wires interconnecting things for any given transport layer.

    Email is just a particular application that requires an interconnected data network to operate. Users benefit when all email operates over a single network, but there may be any number of clients and servers.

    The email protocols (IMAP, POP, SMTP) do comprise a natural monopoly. We are all better off with one set of such protocols. Luckily, these protocols were all put in the public domain before the business community realized how valuable they would be.

    If you look at the “stack” involved in providing a service, you’ll find some levels work best if there is just one of something, while at other layers it doesn’t matter. For example, it is nice that most roads are interconnected so one can drive almost anywhere. It is nice that everyone drives on the same side of the road, at least within a given region. In contrast, the details of each vehicle make little difference, as long as the vehicle fits the lanes and meets certain performance and safety standards.

  9. This is a nice analysis, Kaleberg. I think it’s clear that this issue needs a bit more thought, and that it would help to work through some of the analogies in more detail.

  10. It probably makes sense (probably) to have just *one* network… but who the heck ever said that the various disparate components of said network needs must be operated by one entity?

    Maybe you mean that, in an abstract sense the network *itself* is a monopoly in that there is only one network. Or–if there were at some point multiple separate networks, then when they all join together they form together to make just one larger whole. But if so that usage is mostly not supported by the google definition:
    “the exclusive possession or control of the supply or trade in a commodity or service.”

    The Internet is One Big Network. But it’s not wholly owned or controlled by just one entity.

    In fact, even the road system is not. Within any given governmental region, often the roads are owned and operated by the government. But when you cross a border, some other government controls that section.

    I dunno… maybe the moment the word “monopoly” is taken outside of the purely financial entity context, maybe it becomes too vague to form any worthwhile debate around it.

    TMBG says “There’s Only One Everything”. The Fringe (tv show) says “There’s More Than One Everything”. Both are correct depending on how you define “everything”. Same with “monopoly”.

    I want Board Walk and Park Place.

    Furry cows moo and decompress.

  11. (Sorry for spamming, but) I still take issue with Godin’s original point. Specifically the quote here: “A natural monopoly is a business that benefits its users by being the one and only. If there were two incompatible phone networks”

    He made a totally unwarranted assumption in his “if”. He says “if there were two incompatible…” Fine. But what if there were –>two or more phone businesses sharing a *compatible* network instead who ever said they had to compete by having incompatible networks? <–? They could compete in other ways. It seems like he's mixing up "business" with "infrastructure".

    I don't like Godin.

  12. I agree there was some fuzziness in this particular Godin article — which of course is why I brought it up in the first place. But I very much like Godin in general. I certainly couldn’t keep up anything like the sheer frequency of insights that he manages.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.