The best comment on Twitter and architecture I’ve seen

June 1, 2008 By

It’s the comment left by Michael Kowalchik, aka “MikePK” in response to Matthew Ingram’s post about Twitter’s architecture (or the lack thereof). He’s the CTO of Grazr and makes an important point that every entrepreneur should read. So should every pundit who is giving Twitter crap about being down right now. It is the most important comment I’ve seen in weeks in another blog.

This one comment made me look at Grazr yet again. In the comment Mike seemed disappointed about why the market didn’t show up to enjoy his great architecture. Got me thinking about why Grazr doesn’t have many users and, therefore, doesn’t have Twitter’s scaling problems. Either way, read the comment that Michael left over on Ingram’s blog. The rest of this is just a rant, with a bonus rant about why FriendFeed isn’t going to be Twitter either.

+++++++++++++++++++++++

Here’s why Grazr is no Twitter:

1. Grazr’s name sucks. I HATE HATE HATE “Flickr” copy names. Er, Web 2.0 names. It’s so hard to tell other people about things when you introduce misspellings into them. Here, what’s easier to tell someone else about “FriendFeed” or “Grazr.”
2. Grazr solves a problem normal people don’t have. I think Dare Obasanjo is right, too many companies are trying to solve a problem only the weirdos in society (like me) are having. I explained this on the Gillmor Gang on Friday: I’m a noise junkie. Only one out of 100,000,000 people will be like me. If you think you can build a business just on those weirdos like me or Mike Arrington or Louis Gray will ever use, then go for it. But you don’t need an enterprise-level architecture to keep the two of us happy. Look at Grazr: how many people have too many feeds or want access to more? Only a very small percentage. Who wants to tell their friends what they are eating for lunch? A whole lot more people.
3. Grazr’s UI is too confusing. Look at all the hottest services lately. They are simple, simple, simple. Easy to get into and easy to use. Way too much use of color, too. Why? Put this sucker in front of an eye tracking research project and you’ll see why: you don’t know where to look so your eye gets confused and when it does that the next thing that happens is I look for the “back” button to get the hell out of there.
4. Grazr has a focus on A-list blogs. Who wants to read those things? I’d rather read the blogs from my friends. Those A-list assh***s? I already see too much of them in other places.
5. Grazr’s language is cold. No personality. At least Twitter has the “Fail Whale” with lots of little birds. It has a personality. Grazr? Look at the terms they use for their categories. Business. Celebrity. Gaming. Health. Music. Yahh, yahhh, yahhh, boring!
6. Nothing is moving on Grazr’s home page. I’ve been staring at this for five minutes and nothing has moved. Compare to Twitter Vision — which is more inviting? I even refreshed and nothing on the home page changed. Now go to Twitter or FriendFeed or Jaiku or Pownce. Click on the “everyone” feeds on FriendFeed. Refresh. Refresh. Refresh. Refresh. Refresh. Do you see new stuff? I do. It makes me feel like something is happening on those services and that there’s tons of users. Oh, wait, there are.
7. Grazr has UI that looks like Microsoft’s Windows. Enough said. I know what they are trying to do, but look at FriendFeed’s widget on my blog. Does it look like Windows? No, it’s customized so it fits into my blog’s design.

But, go back to the comment that Michael left. That’s exactly true. I’d rather have Twitter with all of its scalability troubles than a perfect system without any users.

END GRAZR RANT, START FRIENDFEED RANT

That’s why we’re all staying with Twitter. Now, if someone can figure out how to build a perfect system AND get the users to move, then we’ll talk again. FriendFeed is close, but isn’t going to be it. Why? Four reasons:

1. No realtime yet. When I can participate in FriendFeed by using an instant messaging client like Google Talk, then we’ll have realtime. Right now it’s pseudo real time and not wholly satisfying.
2. No SMS compatibility. Can I post to FriendFeed and get messages out of FriendFeed via a cell phone’s SMS feature? Not yet. How many cell phones are being sold everyday? In China alone they are selling six million new ones a month! Now THAT is a market Dare Obasanjo could get excited about!
3. No ability to see a river of noise. Everything on FriendFeed gets reordered based on participation. I want to see just a strict reverse-chronological view.
4. Poor querying abilities. I can’t tell the search to just show me every item that has “n” likes. For instance, I want to see only the popular items sometimes. I can’t do that. Same with comments. I want to see only those items that have lots of community engagement. I can’t. Steve Gillmor asks for this feature another way: he loved Twitter’s track feature. I can’t do that in FriendFeed either.

Oh, well, I’m off on a FriendFeed rant. Enough of that. Thanks Michael for making me think in a different way. What a great comment.

Filed Under: technology industry Tagged With: , , , , ,
← Older Comments
Newer Comments →
  • http://www.inkswig.com Dawn

    Robert, I meant to ask you… Gillmore in that podcast kept saying that you are an investor in FriendFeed, which surprised me since you haven’t told us that here. But then he finally made it sound like you are an investor in time and passion and that he really didn’t mean money.

    So just to clear the air, you haven’t given them money, have you? Not that I would mind, it just seems not to conform to your usual openness, etc., unless I’ve missed a post somewhere.

  • http://www.inkswig.com/ Dawn

    Robert, I meant to ask you… Gillmore in that podcast kept saying that you are an investor in FriendFeed, which surprised me since you haven’t told us that here. But then he finally made it sound like you are an investor in time and passion and that he really didn’t mean money.

    So just to clear the air, you haven’t given them money, have you? Not that I would mind, it just seems not to conform to your usual openness, etc., unless I’ve missed a post somewhere.

  • http://www.inkswig.com/ Dawn

    Robert, I meant to ask you… Gillmore in that podcast kept saying that you are an investor in FriendFeed, which surprised me since you haven’t told us that here. But then he finally made it sound like you are an investor in time and passion and that he really didn’t mean money.

    So just to clear the air, you haven’t given them money, have you? Not that I would mind, it just seems not to conform to your usual openness, etc., unless I’ve missed a post somewhere.

  • http://mikepk.com/ mikrpk

    Tom: I’m confused what your point is, exactly. I never said it was an either/or, I said it was a balance, a balance of focus and energy. “Being on the wrong track” requires energy to explore the right track, it’s not something that occurs for free. There’s a myth that startups are formed from whole cloth with a plan and a path and execute and reach success… the truth is usually a lot muddier, exploring the space you begin in until you find the compelling problem/niche/technology that creates the passionate users. It’s still effort.

    As engineers we obsess over the hard technology problems (scaling) and can overlook other issues. Twitter has the passionate users already, I contend that’s actually a *harder* problem then building a scaling infrastructure.

  • http://mikepk.com/ mikrpk

    Tom: I’m confused what your point is, exactly. I never said it was an either/or, I said it was a balance, a balance of focus and energy. “Being on the wrong track” requires energy to explore the right track, it’s not something that occurs for free. There’s a myth that startups are formed from whole cloth with a plan and a path and execute and reach success… the truth is usually a lot muddier, exploring the space you begin in until you find the compelling problem/niche/technology that creates the passionate users. It’s still effort.

    As engineers we obsess over the hard technology problems (scaling) and can overlook other issues. Twitter has the passionate users already, I contend that’s actually a *harder* problem then building a scaling infrastructure.

  • http://mikepk.com mikrpk

    Tom: I’m confused what your point is, exactly. I never said it was an either/or, I said it was a balance, a balance of focus and energy. “Being on the wrong track” requires energy to explore the right track, it’s not something that occurs for free. There’s a myth that startups are formed from whole cloth with a plan and a path and execute and reach success… the truth is usually a lot muddier, exploring the space you begin in until you find the compelling problem/niche/technology that creates the passionate users. It’s still effort.

    As engineers we obsess over the hard technology problems (scaling) and can overlook other issues. Twitter has the passionate users already, I contend that’s actually a *harder* problem then building a scaling infrastructure.

  • Bob Ngu

    Scoble, do you get stock options every time you mention FriendFeed? :)

    As for MikePK’s comment, I couldn’t agree more. Too many startups spend way too much resources and effort in architecting a highly scalable system only to ask the question “Where are my users?”. This is a trap that technologist often falls into. Better technology does not equal successful startup. It always starts with simplicity and clear value proposition to the mainstream users. For every Twitter, there are probably hundreds, if not more, of highly scalable over-architected sites that went nowhere.

    Tip to Grazr, you already lost all your mainstream users with the term OPML as the first item for feed on your home page.

  • Bob Ngu

    Scoble, do you get stock options every time you mention FriendFeed? :)

    As for MikePK’s comment, I couldn’t agree more. Too many startups spend way too much resources and effort in architecting a highly scalable system only to ask the question “Where are my users?”. This is a trap that technologist often falls into. Better technology does not equal successful startup. It always starts with simplicity and clear value proposition to the mainstream users. For every Twitter, there are probably hundreds, if not more, of highly scalable over-architected sites that went nowhere.

    Tip to Grazr, you already lost all your mainstream users with the term OPML as the first item for feed on your home page.

  • Bob Ngu

    Scoble, do you get stock options every time you mention FriendFeed? :)

    As for MikePK’s comment, I couldn’t agree more. Too many startups spend way too much resources and effort in architecting a highly scalable system only to ask the question “Where are my users?”. This is a trap that technologist often falls into. Better technology does not equal successful startup. It always starts with simplicity and clear value proposition to the mainstream users. For every Twitter, there are probably hundreds, if not more, of highly scalable over-architected sites that went nowhere.

    Tip to Grazr, you already lost all your mainstream users with the term OPML as the first item for feed on your home page.

  • http://mikepk.com/ mikepk

    Darn typo… supposed to be mikepk, not mikrpk. :)

  • http://mikepk.com/ mikepk

    Darn typo… supposed to be mikepk, not mikrpk. :)

  • http://mikepk.com mikepk

    Darn typo… supposed to be mikepk, not mikrpk. :)

  • http://scripting.wordpress.com/ Dave Winer

    River of Noise — I love it!

    That’s beautiful.

  • http://scripting.wordpress.com/ Dave Winer

    River of Noise — I love it!

    That’s beautiful.

  • http://scripting.wordpress.com/ Dave Winer

    River of Noise — I love it!

    That’s beautiful.

  • http://scripting.wordpress.com/ Dave Winer

    Here’s the opposite of River of Noise.

  • http://scripting.wordpress.com/ Dave Winer

    Here’s the opposite of River of Noise.

  • http://scripting.wordpress.com/ Dave Winer

    Here’s the opposite of River of Noise.

  • http://scobleizer.com/ Robert Scoble

    I invest my time, not my money in these services.

  • http://scobleizer.com/ Robert Scoble

    I invest my time, not my money in these services.

  • http://scobleizer.com/ Robert Scoble

    I invest my time, not my money in these services.

  • http://scobleizer.com/ Robert Scoble

    Bob: I get nothing from these services. If that ever changes I will disclose it.

  • http://scobleizer.com/ Robert Scoble

    Bob: I get nothing from these services. If that ever changes I will disclose it.

  • http://scobleizer.com/ Robert Scoble

    Bob: I get nothing from these services. If that ever changes I will disclose it.

  • http://scobleizer.com/ Robert Scoble

    Tom: if you are an entrepreneur and you only hire technical resources then you focused too much of your energy on one thing, when a business needs more than just a great technology to succeed. This is a resource problem. OK, maybe it’s not the CTO’s problem to solve, but it’s SOMEBODY’s job to solve, otherwise the company won’t get the attention, er, usage/customers, it deserves.

  • http://scobleizer.com/ Robert Scoble

    Tom: if you are an entrepreneur and you only hire technical resources then you focused too much of your energy on one thing, when a business needs more than just a great technology to succeed. This is a resource problem. OK, maybe it’s not the CTO’s problem to solve, but it’s SOMEBODY’s job to solve, otherwise the company won’t get the attention, er, usage/customers, it deserves.

  • http://scobleizer.com/ Robert Scoble

    Tom: if you are an entrepreneur and you only hire technical resources then you focused too much of your energy on one thing, when a business needs more than just a great technology to succeed. This is a resource problem. OK, maybe it’s not the CTO’s problem to solve, but it’s SOMEBODY’s job to solve, otherwise the company won’t get the attention, er, usage/customers, it deserves.

  • http://friendbinder.com/users/rythie Richard Cunningham

    Robert, I think 1st point and in part the second one too are going to be difficult for lifestreaming sites in general. The reason is that not that the lifestreaming site couldn’t have a Instant messaging interface (e.g. XMPP), it’s that getting this information for the source site (or network) is difficult.

    With RSS and Atom you can get this information quickly because of ping services which notify you when someone posts, if you are willing to get this information for everybody. Also Twitter provides access to it’s public timeline via XMPP. However with other sites you have to poll and they have limits of the number of requests so you can’t poll that often. If that wasn’t bad enough on some sites it takes quite a few requests to just find out what your friends are doing.

    I think in general the source sites’ APIs are not designed for this type of usage and the challenge for them will be to create some sort of push API.

    Not wanting to plug my own site too much but, in response to point 3, friendbinder.com has a “strict reverse-chronological view”. However we only opened up the private beta on Friday and it wasn’t designed for people like you (because as you say there are not many like you!) and I know currently it can’t support more than 1000 twitter users very well.

  • http://friendbinder.com/users/rythie Richard Cunningham

    Robert, I think 1st point and in part the second one too are going to be difficult for lifestreaming sites in general. The reason is that not that the lifestreaming site couldn’t have a Instant messaging interface (e.g. XMPP), it’s that getting this information for the source site (or network) is difficult.

    With RSS and Atom you can get this information quickly because of ping services which notify you when someone posts, if you are willing to get this information for everybody. Also Twitter provides access to it’s public timeline via XMPP. However with other sites you have to poll and they have limits of the number of requests so you can’t poll that often. If that wasn’t bad enough on some sites it takes quite a few requests to just find out what your friends are doing.

    I think in general the source sites’ APIs are not designed for this type of usage and the challenge for them will be to create some sort of push API.

    Not wanting to plug my own site too much but, in response to point 3, friendbinder.com has a “strict reverse-chronological view”. However we only opened up the private beta on Friday and it wasn’t designed for people like you (because as you say there are not many like you!) and I know currently it can’t support more than 1000 twitter users very well.

  • http://friendbinder.com/users/rythie Richard Cunningham

    Robert, I think 1st point and in part the second one too are going to be difficult for lifestreaming sites in general. The reason is that not that the lifestreaming site couldn’t have a Instant messaging interface (e.g. XMPP), it’s that getting this information for the source site (or network) is difficult.

    With RSS and Atom you can get this information quickly because of ping services which notify you when someone posts, if you are willing to get this information for everybody. Also Twitter provides access to it’s public timeline via XMPP. However with other sites you have to poll and they have limits of the number of requests so you can’t poll that often. If that wasn’t bad enough on some sites it takes quite a few requests to just find out what your friends are doing.

    I think in general the source sites’ APIs are not designed for this type of usage and the challenge for them will be to create some sort of push API.

    Not wanting to plug my own site too much but, in response to point 3, friendbinder.com has a “strict reverse-chronological view”. However we only opened up the private beta on Friday and it wasn’t designed for people like you (because as you say there are not many like you!) and I know currently it can’t support more than 1000 twitter users very well.

  • http://avc.blogs.com/ fred wilson

    Robert

    I don’t think anyone wants FriendFeed to be Twitter 2.0

    We want FriendFeed to be the best social media aggregation system. Which it already is.

    I think we all win (the users) if the best of breed services keep getting better at the one thing they do better than anyone else.

    Fred

  • http://avc.blogs.com/ fred wilson

    Robert

    I don’t think anyone wants FriendFeed to be Twitter 2.0

    We want FriendFeed to be the best social media aggregation system. Which it already is.

    I think we all win (the users) if the best of breed services keep getting better at the one thing they do better than anyone else.

    Fred

  • http://avc.blogs.com fred wilson

    Robert

    I don’t think anyone wants FriendFeed to be Twitter 2.0

    We want FriendFeed to be the best social media aggregation system. Which it already is.

    I think we all win (the users) if the best of breed services keep getting better at the one thing they do better than anyone else.

    Fred

  • http://scobleizer.com/ Robert Scoble

    >I don’t think anyone wants FriendFeed to be Twitter 2.0.

    Oh, Steve Gillmor wants that! :-)

    I agree that it’s the best social media aggregation system. It’s also the best conversation system on top of that aggregation. But, it could be so much more.

    It’s “how much more” that we’re arguing over here.

  • http://scobleizer.com/ Robert Scoble

    >I don’t think anyone wants FriendFeed to be Twitter 2.0.

    Oh, Steve Gillmor wants that! :-)

    I agree that it’s the best social media aggregation system. It’s also the best conversation system on top of that aggregation. But, it could be so much more.

    It’s “how much more” that we’re arguing over here.

  • http://scobleizer.com/ Robert Scoble

    >I don’t think anyone wants FriendFeed to be Twitter 2.0.

    Oh, Steve Gillmor wants that! :-)

    I agree that it’s the best social media aggregation system. It’s also the best conversation system on top of that aggregation. But, it could be so much more.

    It’s “how much more” that we’re arguing over here.

  • Pingback: Gemilo blog » Blog Archive » Sosiaalisen median kehittäjien muistilista

  • http://mikepk.com mikepk

    There is a tyranny of the feature though you need to be careful of. We built so many features into Grazr that I can barely keep track of them all anymore. I’m constantly running into things, “we do that? oh yeah I forgot we built that feature”. You can build so many features that the character of the product changes and it becomes no longer accessible to the newcomer. It’s again one of those balances, between power users and first time experience.

    Part of the reason we built so many features was the hope that someone would find a subset of them unique and useful and we could allocate resources to better satisfy those individuals needs. The serious downside though is that if you’re not careful it can negatively impact your UI and usability. More importantly though it can negatively affect your message “what *are* you?”. That’s a question we struggle with.

    Friendfeed does have an advantage that they’ve waited for people to digest their current feature set before adding. I think one of our issues is that we steamrolled ahead a little too quickly in expanding the service. We’re actually considering stripping out a lot of the functionality to deliver a clearer message.

  • http://mikepk.com/ mikepk

    There is a tyranny of the feature though you need to be careful of. We built so many features into Grazr that I can barely keep track of them all anymore. I’m constantly running into things, “we do that? oh yeah I forgot we built that feature”. You can build so many features that the character of the product changes and it becomes no longer accessible to the newcomer. It’s again one of those balances, between power users and first time experience.

    Part of the reason we built so many features was the hope that someone would find a subset of them unique and useful and we could allocate resources to better satisfy those individuals needs. The serious downside though is that if you’re not careful it can negatively impact your UI and usability. More importantly though it can negatively affect your message “what *are* you?”. That’s a question we struggle with.

    Friendfeed does have an advantage that they’ve waited for people to digest their current feature set before adding. I think one of our issues is that we steamrolled ahead a little too quickly in expanding the service. We’re actually considering stripping out a lot of the functionality to deliver a clearer message.

  • http://mikepk.com/ mikepk

    There is a tyranny of the feature though you need to be careful of. We built so many features into Grazr that I can barely keep track of them all anymore. I’m constantly running into things, “we do that? oh yeah I forgot we built that feature”. You can build so many features that the character of the product changes and it becomes no longer accessible to the newcomer. It’s again one of those balances, between power users and first time experience.

    Part of the reason we built so many features was the hope that someone would find a subset of them unique and useful and we could allocate resources to better satisfy those individuals needs. The serious downside though is that if you’re not careful it can negatively impact your UI and usability. More importantly though it can negatively affect your message “what *are* you?”. That’s a question we struggle with.

    Friendfeed does have an advantage that they’ve waited for people to digest their current feature set before adding. I think one of our issues is that we steamrolled ahead a little too quickly in expanding the service. We’re actually considering stripping out a lot of the functionality to deliver a clearer message.

  • Bob Ngu

    @Scoble, I was half-kidding anyway but thanks for clarifying, good to know =)

    @mikepk, agreed, that’s another thing us technologist have to watch out for, unending feature creep. IMO, for Web 2.0 sites, it’s much better to get the features out earlier than later because users’ feedback will help drive the changes. Often times, we can’t predict what the users needs are and ended up all over the map with a plethora of features that aren’t needed if we wait too long to get it in front of them.

    Like you, I have learned that too many features confused the users and caused them to lose interests. Also, too many features can make the site seems like a horizontal play and that’s very hard (IMO) these days to build mass traction.

    Just my $0.02.

  • Bob Ngu

    @Scoble, I was half-kidding anyway but thanks for clarifying, good to know =)

    @mikepk, agreed, that’s another thing us technologist have to watch out for, unending feature creep. IMO, for Web 2.0 sites, it’s much better to get the features out earlier than later because users’ feedback will help drive the changes. Often times, we can’t predict what the users needs are and ended up all over the map with a plethora of features that aren’t needed if we wait too long to get it in front of them.

    Like you, I have learned that too many features confused the users and caused them to lose interests. Also, too many features can make the site seems like a horizontal play and that’s very hard (IMO) these days to build mass traction.

    Just my $0.02.

  • Bob Ngu

    @Scoble, I was half-kidding anyway but thanks for clarifying, good to know =)

    @mikepk, agreed, that’s another thing us technologist have to watch out for, unending feature creep. IMO, for Web 2.0 sites, it’s much better to get the features out earlier than later because users’ feedback will help drive the changes. Often times, we can’t predict what the users needs are and ended up all over the map with a plethora of features that aren’t needed if we wait too long to get it in front of them.

    Like you, I have learned that too many features confused the users and caused them to lose interests. Also, too many features can make the site seems like a horizontal play and that’s very hard (IMO) these days to build mass traction.

    Just my $0.02.

  • Pingback: Everything I Know About Twitter I Learned on Prom Night « tchblg

  • http://www.softwaresoapbox.com/ Brian Benz

    Huh. what a name! Should have called it impostr. If you get too much content, maybe pestr.

  • http://www.softwaresoapbox.com/ Brian Benz

    Huh. what a name! Should have called it impostr. If you get too much content, maybe pestr.

  • http://www.softwaresoapbox.com Brian Benz

    Huh. what a name! Should have called it impostr. If you get too much content, maybe pestr.

  • Scobby

    ARHHGHG … Scoble slow the hell down. You are changing positions faster than a politician during a the heat of a campaign. My head is starting to hurt. I can´t keep up …

  • Scobby

    ARHHGHG … Scoble slow the hell down. You are changing positions faster than a politician during a the heat of a campaign. My head is starting to hurt. I can´t keep up …

  • Scobby

    ARHHGHG … Scoble slow the hell down. You are changing positions faster than a politician during a the heat of a campaign. My head is starting to hurt. I can´t keep up …

← Older Comments
Newer Comments →