Twitter is Like the Weather

Warning: This post is about Twitter, so if you don’t care about Twitter, stop reading here. Proceed at your own risk of boredom.

Last week, friends of the ‘Lab David Haimes and Michael Krigsman were exchanging some jabs over Twitter, about Twitter. Michael argues that Twitter’s frequent outages make it appear suspect if/when a business model emerges.


David counters that the business model should be the focus now, not bulletproof reliability.


This is a chicken/egg problem. Was the service inherently unreliable, or did the massive amount of freeloading traffic cripple it?

Over the past few weeks, Twitter has been especially unreliable, which for me is mildly annoying, but nothing that upsetting. Michael blogged several posts about Twitter as an IT failure, which is in his wheelhouse, and the echo chamber was full of coverage about how it might be Scoble’s fault, how Twitter was poorly architected for its current form, how FriendFeed should kill Twitter, etc.

In light of the frequent downtime, it was nice of Twitter to stay up for David and Michael to disagree about it. David’s response, blogged in paragraphs of 140 characters, a nice touch, essentially sums up how I feel about the whole thing. So what if Twitter’s down?

Twitter is free; you get what you pay for; depending on a free service seems like a slippery slope, since there are no guarantees for uptime. People generally refer to uptime in terms of the number of 9s, usually guaranteed in the SLA, e.g.

99.9% uptime (”3 nines”) is 21.9 hours downtime/year
99.99% uptime (”4 nines”) is 52.56 minutes/year
99.999% uptime (”5 nines”) is 5.26 minutes/year
99.9999% uptime (”6 nines”) is 31.54 seconds/year

These are tough to maintain for profit, so why would you try for free?

Of course, Twitter doesn’t have an SLA because it’s a free service, and good thing, because the last report I saw had their uptime at only one 9, 98.72%. That number is sure to have fallen after the dark days of May.

When Twitter’s down, people get up in arms. Then, when it’s back and humming along, people act like nothing happened. It’s like having a fight with a loved one, followed by a tearful, “let’s never fight again” makeup. Even in the absence of a business model and in the absence of service a lot of the time, people have not abandoned Twitter, even for clones like Pownce and Jaiku. Why?

For me, it’s general apathy about the service (see above), but more importantly, Twitter has the network, the first-mover status. This is a big barrier to switching for me, and since I don’t find a large utility in Twitter to begin with, I’m generally fine living through the dark times.

Maybe over time, my utility will rise, which I’m guessing is what Twitter is banking on; get your users so entrenched that they are willing to stomach whatever revenue model you put in place.

Or maybe they’re making it up as they go. Either way, it’s fun to watch. I’ll leave you with this hilarious take on what Twitter’s business model should be.

Update: Nick O’Neill ponders whether reliablity matters to social media, with Twitter as a prime example. He notes that according to Compete, people don’t seem to care about Twitter’s reliability, as evidenced by its continued brisk growth. Twitter Unique Visits




  1. Twitter has a large, loyal and passionate user base. People can't see a micro-blogging world where Twitter isn't dominant. But then again, years ago, I couldn't see a search engine coming along to displace AltaVista.

    [Jaiku launched 6 months before Twitter so it's slightly unfair to label them as a Twitter 'clone'.]

  2. Passionate being the keyword. Power balance in the Intertubes does shift, remember Excite?

    Regarding history, Jaiku formed a company in 2/06 and launched its service in 8/06 in beta. It left beta in 3/07. Twitter started within Obvious in 3/06 and launched its service in 10/06. Some would argue it's still in beta 🙂

    Their history is pretty closely tied, but even a year ago, Jaiku was seen as a clone. So, I guess perception is reality.

  3. I find it interesting that two companies in different parts of the world were moving in exactly the same direction concurrently.

    Too bad we can't see an architecture bake-off between the two. From a user's view, they seem very similar, but somehow I think the architecture is very different.

    I suspect Google is watching Twitter's very public travails with interest. I expect them to integrate Jaiku soon b/c let's face it Google/Jaiku can bring micro-blogging to a larger user base that doesn't know about Twitter.

  4. Will you use Google Search if you were ask to pay for it? Twitter is popular because its free. They should look at attracting more customers now and increasing reliability. With the current image, building a business model over something that is unreliable would be difficult or rather impossible.

  5. Hard to attract new users in the face of shoddy reliability, as Michael says, but most of us are hooked and willing to ride out the dark times.

    Paying for a service isn't the only business model online; in fact, it's probably the least common, due to the free nature of the Intertubes and the fact that there are plenty of clones.

    Twitter is not only popular because it is free; the service and network make it valuable to users. Pownce and Jaiku are also free, but they are not nearly as popular asTwitter. Lots of new, free services launch every month, but they're not all popular just b/c they're free. Free is expected.

  6. A lot of people send and receive tweets via SMS, so I wonder if they could get some share of the revenue from that and with some tight integration into the mobile provider phonesvand services they could accelerate adoption too. In the UK where local phone calls are not free, many dial up ISP would offer a free service andmake money taking a share of the phone revenue.

    If anyone at Twitter reads this and haven't thought of it yet thought of it – contact me to work out how much you're going to pay me for the idea.

  7. I am out of the office until Monday, June 9th and will have limited access to email during this time.

    You can reach me on my cell (415 269 3616) for urgent issues, or contact my manager Rob Zwiebach. Otherwise I will contact you on my return.


  8. I am out of the office until Monday, June 9th and will have limited access to email during this time.

    You can reach me on my cell (415 269 3616) for urgent issues, or contact my manager Rob Zwiebach. Otherwise I will contact you on my return.


  9. Sure, David has a point that Twitter needs a business model. But, that's not what it needs today. With $20M in funding, Twitter has all the funding it requires for the foreseeable short-term future. Right now the company needs to manage it's infrastructure with the goal of utility-level reliability. If they don't accomplish this, Twitter will die.

    Note to David: Dead companies don't need a business model.

    Michael Krigsman

  10. Agreed, Twitter has many possible business models from which to choose. It would be a hoot if they crowdsourced it too. That would ensure a palatable model that would appeal to the majority of users.

  11. I still say this whole debate is chicken/egg. What's the point in achieving ironclad reliability if people are willing to forgive Twitter? It's become a joke among users, and I've still not seen an exodus. Have you?

    On the positive side, maybe the flakiness means people are less likely to base their own businesses on Twitter, which is a good thing, since Twitter is a free service. So, what's wrong with “good enough” reliability?

    I hope the VCs who ponied up the most recent round kicked the business model tires too.

  12. Notes to Michael:
    1) Dead companies don't need to manage infrastructure.
    2) $20m does not go far, especially when trying get 3 or more 9s availability.
    3) Many companies who manged infrastructure well are dead, due to a lack of business mode, webvan for example.


  13. Jake,

    You would assume the VCs who kicked in some cash have some idea how to make money out of this longer term. However I have seen VC money thrown at some crazy stuff with no chance of ever making money. Hindsight is always 20/20 😉

  14. I do find it funny (and possibly smart) that Twitter dialed down expectations and some features in advance of Apple WWDC. Reminds me of that old SNL skit for the dating service called “Lowered Expectations”.

    Ah, Twitter. Sigh.

Leave a Reply

Your email address will not be published.

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