-
Website
http://www.scobleizer.com/ -
Original page
http://scobleizer.com/2008/05/30/twitter-blames-its-users/ -
Subscribe
All Comments -
Community
-
Top Commenters
-
danja
44 comments · 4 points
-
polizeros
52 comments · 1 points
-
AndyBeard
69 comments · 4 points
-
Zachary Adam Cohen
35 comments · 8 points
-
dbarefoot
40 comments · 3 points
-
-
Popular Threads
-
The best and worst thing Twitter did in 2009: RT
16 hours ago · 20 comments
-
World-brand-building mistakes France’s entrepreneurs make
1 week ago · 181 comments
-
2010: the year SEO isn’t important anymore
6 days ago · 66 comments
-
iPhone developers abandoning app model for HTML5?
6 days ago · 51 comments
-
A 2010 real-time app development platform from Kynetx
14 hours ago · 2 comments
-
The best and worst thing Twitter did in 2009: RT
So, no, even this headline isn't accurate.
""The events that hit our system the hardest are generally when “popular” users - that is, users with large numbers of followers and people they’re following - perform a number of actions in rapid succession.""
That sure sounds like blaming their best users, and very specifically, me, since I'm the only user with more than 20,000 followers that behaves this way.
If everyone in NYC flushes the toilet at the same time the sewer system will crash.
If everyone picks up their phone at the same time many won't get dial tone.
Are these bad services?
Systems have design parameters. You may be using the twitter system outside of it's engineered parameters. If the system is designed for "friends" and not "followers" then 25,000 is not a reasonable number. What % of your followers are friends?
I have 9,000 business cards. Are they not my friends? How do you define friends online? I define them as someone who I want in my social network.
At least Facebook just limits us to 5,000.
FriendFeed, though, already has me up to 11,556 and works great.
"This usually results in a number of big queries that pile up in our database(s). Not running scripts to follow thousands of users at a time would be a help, but that's behavior we have to limit on our side."
Note that last bit: Alex is saying that it's a problem that's on THEIR side, not something that users can fix. They're not asking for a change in behaviour: they're giving s description of how their system behaves under stress.
there is a HUGE difference.
"Our key problems have been primarily architectural and growing our infrastructure to keep up with our growth. "
"We're painfully aware of every minute that we're slow or unavailable."
"Rails excels at CRUD-style applications, and while you can wedge a messaging system or other types of applications into that model, it's a square peg in a round hole."
"My understanding is that Twitter started as a one-day project to explore sharing status via SMS that rapidly took on a life of its own. That Twitter would eventually evolve into a messaging system in its own right wasn't conceptualized from the get-go."
And there's more. Once you put Alex's quote into context, it is VERY clear that he's not blaming the users for anything - he's explaining the limitations of the architecture.
I think that Investor are crazy to with 15M$ in the actually Twitter sistem: "Rails was originally extracted from Basecamp, a CMS. Rails excels at CRUD-style applications, and while you can wedge a messaging system or other types of applications into that model, it's a square peg in a round hole."
So, i'am Italian, yesterday i talk about Twitter with my friend: she doesn't understand why she has to use Twitter when she can use E-mail, SMS, Msn Messenger in order to contact friends. I think that is a pretty question.
Robert, of course your behaviour isn't causing it. No one is saying that it is. You're reading into this something that just isn't there.
Ouch Scoble that hurts, you say you are their best user just because you use it a lot? I must be one of the worst users then since I have only a few friends and only a few posts.
I guess that means spam emailers are the best emailers because they send the most mail.
As noted earlier, the comment from twitter is stating a fact that their system is hurt the most by heavy users, I don't call that blaming, I call that plain and simple fact that can be proven by data. They are working it, they have said this several times now.
The reporter asks: "if there’s anything users can do to lighten our load."
Twitter responded by saying, yes, our heavy users need to back off for a bit while we work on the issue.
Headlines are wrong, the report used your name not twitter.
Calm down already.
We should have a contest to see who can program the best replacement. It shouldn't be too hard to generate a synthetic load of 10,000 tweets/second for a baseline.
--Mike--
This is all about contention based systems. It's about peak to average ratios.
This is like a B$M retailer telling its customers to go away because the weight is cracking the foundation. Better yet, telling its overweight customers to go away. If it cant handle the load then shut down, rebuild, and have a grand reopening. Dont have strange hours and keep the customers guessing as to when you'll be open.
"charles asks if there's anything users can do to lighten our load. The events that hit our system the hardest are generally when "popular" users - that is, users with large numbers of followers and people they're following - perform a number of actions in rapid succession. This usually results in a number of big queries that pile up in our database(s). Not running scripts to follow thousands of users at a time would be a help, but that's behavior we have to limit on our side."
http://dev.twitter.com/2008/05/youve-got-qs-wev...
If you really want to read into it, what they are saying is they have scalability problems, and actions that have a lot of side effects - not limited to the 'big' users - stress the system. Are you surprised? They are not blaming anyone, rather answering what users can do to help if they choose to.
You may be the biggest user, I don't know, but please don't bash Twitter when they actually start opening up, especially when they haven't 'blamed' you!
BigCheeze, I see signs all the time for weight restrictions on trucks on certain roads, so why can't their be for an online service?
Are you willing to man up to being wrong?
Bare in mind, too, that they're constantly working on the system - as they put it, replacing the bits which won't scale with bits which will. Maintainance can always cause downtime which has little to do with user traffic. Saying "it's down, and I'm not tweeting, so therefore high-numbers users can't be a problem" is silly.
Seems clear enough to me.
I've read such incredible things about Evan, so that doesn't make any sense.
Twitter has been around for a while now, and yet they don't seem to have figured out how to enable people to use their service. Blaming people for trying to use the service isn't going to help matters. Do they want users or not?
What makes it worse is that Robert has been a great evangelist for the service for a long time. His posts made me join, and probably encouraged lots of others too. That makes any insult (real or supposed) cut deeper.
This pisses me off because I'm a software engineer and you sound like every idiot who has a complaint and thinks it should be fixable in ten minutes. Chill out and let them work on it.
Friendfeed doesn't have the same traffic that Twitter does so your comparisons aren't valid. The scale counts. I would be surprised if the top users on FF follow the same number of people they did on Twitter (and you alone as a datapoint doesn't count - you're just one user among thousands).
Saying that you're their only popular user is assigning a wee-bitt too much importance to yourself, IMHO.
Twitter's post was nothing more than them trying to explain what's causing them problems. You've pulled their quotes out of context and written a highly inaccurate headline. They're blaming no one but themselves
Turning things off may have been just as damaging, maybe they didn't have proper ways to turn things off in a good way. etc. etc.
Scoble will never be happy, hence his jumping around from application to application often times because of a few minor hiccups that ALL software and piece of anything is going to have. Yes even Apple :)
It seems like Twitter simply needs to grok two words: asynchronous and queuing
Get over yourself... the world doesn't revolve around you. A week or two ago you and Arrington were bashing Twitter for not communicating well, now you're bashing them for overcommunicating.
I don't work there, but I suspect there are a bunch of twenty somethings that spend 100+ hour weeks trying to keep it running with bandaids. Was it architected correctly from the get go? No. But that probably has more to do with the funding they had than the people they had at the time.
So what if it is your partially due to how you use it? They are just explaining what happened, not saying that you should use it differently.
Instead of manufacturing this fake controversy for attention, why don't you use your podium to start a discussion on what technologies and design patterns might be good candidates to replace the failing architecture.
The blogosphere came to prominence because the mainstream press couldn't just competed for the lowest common denominator of intelligence offering sex, violence, and other brain numbing content. Now I see the blogosphere doing the same. And for that, Scoble, you are certainly one to blame.
You draw a comparison with a completely different type of use:
> funny, American Idol causes millions of phone calls to happen and the system doesn’t crash.
This is a misunderstanding. In the AI call-ins, people are registering a number on a counter with their call. One call = +1. On Twitter, when you ask for an update of tweets, there is a massive pull of information from the database. Imagine the difference between you sitting on a bench and counting people walking by vs. asking each one what they had for breakfast. It's a very big difference and unfortunately makes the comparison nonsensical.
Please re-read the actual post and try to understand that it's not all about you. This hysteria-baiting is only hurting your relationship with people who provide a service that you say you love, but lately all you give it publicly is anything but.
"Not running scripts to follow thousands of users at a time would be a help, but that’s behavior we have to limit on our side."
That is clearly not you - it's people who try to friend everybody in the world using rogue app's like this: http://www.livelybrowser.com/img/TwitterFriendA...
Hopefully Alex from Twitter will do a follow-up post to better explain what he meant by those statements.
I'd prefer Alex spend his time working on twitter and not trying to explain himself about something that isn't even there.
New headline: "Scoble blames Twitter for something Twitter didn't say."
FriendFeed offers more and it will likely scale better, but it too is hardly unique in its technology. I can make more sophisticated aggregators with Yahoo! Pipes and I am not even a programmer. I am hopeful, however, that FriendFeed will build in more functionality so that I don't have to continue using both in tandem.
Give it up Robert. It's not always about you. And even if you were to take all of your kool-aid drinkers to FriendFeed with you, there will still be millions of users on Twitter who believe it or not, have never even heard of Robert Scoble.
As many on this thread have stated in one way or another, we live in a real-time Beta testing word. The benefit is understanding right away the strengths and weaknesses of a product or service. The challenge is not pissing off users to the point where they leave. I think we ask a lot of users these days.
I just wish *I* had 11K friends....not!
;-) LA
Get your panties out of a bunch. They never blamed you. They were identifying difficulties and communicating with their users.
I guess its the advertisers who should be more intelligent. That's my only hope now.
[I just posted this def'n on Twitter]
That would be unfair though, because Twitter isn't blaming top users. They're simply blaming their inability to scale properly, and lamenting the fact that the application needs to be rearchitectured for its new use.
"@" functionality, direct messaging, and all that was not part of the original plan. It was originally going to be like a mini blog system, except posts were limitd to 140 characters and distributed via text message. As the service grew, users reappropriated the service into an ad-hoc messaging platform; the application we use today. The Twitter we now know is a much better, stronger program because of that reappropriation, but it's also the reason for these problems. As Alex said, it's fitting a square peg into a round hole.
You can also continue to gallantly defend FriendFeed and all its virtues (and I do believe they have a strong product worth defending and using) but realize that they were able to plan for this growth in advance. They built this scaling into the application, having seen and understood what Twitter went through. Not to mention the fact that they are still far behind Twitter's traffic, so they haven't had the architecture issues Twitter has faced.
Well, you have a big ego Scoble. Sure, you're probably putting a more noticeable strain on the system than I might, but you're also just one guy. There's only about a trillion spammers on Twitter who are writing scripts to "follow" all-and-sundry in the hopes that they'll get more exposure. Scripts which "perform a number of actions in rapid succession". Sorry, but while you might be part of the equation, I don't think this is a case of Twitter "blaming it's users". More like blaming the spammers.
Facebook doesn't want you.
Twitter doesn't want you.
FriendFeed eventually won't want you.
Perhaps you should give it up.
Last 2 cents: I think we are all a little too eager to 'blame' someone, instead let’s look at what we are learning - Things break. Geeks fix them. Users figure it out. People are narcissists. That is why Twitter is so successful and Scoble is getting so much grief about this post.
Time to move onto the next shiny toy. Another notch in the Scoble Curse, hype up, to kill off, from Tablets/UMPC, to PDC/Longhorn, to Seagate (Sue Sue Sue), to Second Life (the new OS), to Facebook, to Twitter. The smart PR person will understand the template and pitch you to hype up competitors products.
This early adopter rot sure has a short lifespan, but sure fun to watch the all the worms frenzy-wiggle before the microwave-ray of reality baconizes it all.
I'm really sorry but I don't understand the point of this post, it only increases the noise in the blogosphere adding very little value to everyone
Just delete your account !
People will find you !
jp
I've not had an opportunity to open an account on FriendFeed, some things I like, others I do not, but nevertheless a migration is indeed happening and I might be one of them.
I would be much happier if they just took the entire thing down and said "give us one week and we'll work 24/7 to give you the best Twitter we can."
Having the Twitter piece mealed (this works, this does not) is unacceptable. I sometimes forget what works and doesn't and when I attempt to do something it only infuriates me more.
Get your act together people, you're losing fan base...and fast.
Gentlemen, your egos, and self centered delusions that you are the center of the universe, are getting old fast.
As for the phone analogy mentioned earlier, you mean to tell me you've never heard an "all circuits are busy now" message when making a phone call? Sounds like a scalability issue to me.
I agree with your point of view on this.
Anyway, let's see how Friendfeed will scale when more user will join.
Notice that there wasn't a line pointed out - what is 'popular' in this context - 100+ follower/ings? 1000+? 10k+? How frequent is 'too frequent'? 3 tweets a day? and hour? 30?
The finger pointing was there - no matter how much I want to defend Twitter because I really want to see them succeed.
There was a brief reference at the end about script adders - but why not blame those folks from the get-go? Why not say "those accounts mass adding/following"? Because we know who someone is referring to by the word "popular", don't we?
The only reason I didn't make Dave Winer's "spewage" list is because he's never followed me (which is perfectly normal) but the %-following math has me right up there, thanks.
So I'm one of the reasons Twitter is having issues? I think not. Long before I had even 1 follower there were folks with 10k plus updates, thousands of followers and "popularity" - so if that was the issue, it should've been addressed last year some time... But it wasn't.
Look, let's be serious here for a minute. Twitter has had stability issues for a long time. They seemed to remedy a LOT of them around early February this year in time for SXSW. Then right about the time Blain leaves, we start seeing more and more issues. I'm not buying it. Suck it up Twitter and admit that the problem isn't coming from the outside, it's internal.
I am a fan of yours but surprised by this post.
I've been watching your interview with Biz and Ev and you are laughing and giggling. No critical questions, just laughing your nerves away.
And now with your computer at hand and no 'real' persons you start criticising on harsh way .
Be brave! Say those things interviewing !
Hopefully, Twitter will acquire some matured management to guide in issues such as this.
- scott -
How much are they paying for bandwidth? Anybody have a clue?
Or, hell, I might just shut the whole thing down. Again, the TOS says I can do that.
I read your post regarding how to better organize your comments threads.
I am curious to know why you think cocomment is not satisfying your needs.
We show them in order of appearance with a drop down list per comment.
I am curious to know what your needs are.
You can reach me at joaquin at cocomment.com
Tks
J
http://www.abhishekkant.net/2008/05/eggs-in-bas...
Seriously, they've previously stated that they designed twitter badly and are trying to fix it. Nobody is getting blamed, they're just saying that their system does note scale well and the database is getting way too many unnecessary calls because of it. (And hint that heavy users are generating many of them. You know it's true, man. ;-) )As a fellow programmer, I can sympathize that it's not the easiest or quickest thing in the world to fix broken architecture.
Anyway. Anyone else noticed that they have a new status screen for over capacity? Hilarious image. :-)
http://static.twitter.com/images/whale.png
it is beyond me why you are so popular.
so you are offended and feel that it was insinuated by twitter that the big scoble is solely to blame. you take the venture beat post as an excuse to unleash your anger towards twitter when they are admitting plainly that their service is stuck between a rock and a hard place.
they are not making excuses that are bullshit. they are pointing out several causes of problems... one of which has to do with the way popular (nothing to do with best, btw) users are handled... in all aspects. it is a known fact now that active users with many followers require intense processing. this in conjunction with other issues such as 3rd party services, IM messaging and an overall limited info architecture that was admittedly not conceived to be what twitter has become. this has all been stated clearly by twitter and others.... you even.
i commented on dave winer and om malik's blog recently about this topic and expressed that if i were one of these resource intensive users... i would voluntarily take some measures to ease things for twitter.... whether or not they needed or wanted such action by a user of their service.... just because i think they deserve that kind of support from the community who has used the hell out of their service for the past 12-18 months.
they are in a unique position. it seems right to support them as the leading company in the micro-messaging space. because like it or not, a user like you is almost like a dos attack on twitter at this point. and if i was a very active popular user, i would chill out and offer to do whatever to ease the burden until they are able to sort through the priority issues of fixing and re-archtecting components so that twitter can be what it has evolved to be.
you make alot of comparisons to other technologies and services when trying to make a case that twitter sucks. but i dont think any of those points have merit... and most are moot. you can praise friendfeed all day long but it's not a fair comparison. and your other comparisons show your lack of technical understandings on the depths of system architecture. you are tech buzz guy.... one that swims in the surface area.... one that gets drunk on the hype. if you truly understood how all the related tech works... you would not be so hasty with your comments and big attitude. but it must be tough to tame that ego with all those very close friends following your digital ways.
get over yourself. i sense that significant others are getting over you by now. i bet a good portion of your "friends" are just in it for the humor and entertainment and mockery aspect. because you dont have much to offer when it comes down to it. you just happen to be popular. and popular is never exclusive to the "best". look around and you will see that to be true... very true.
cheers.
sull
Josh Chandler
http://www.joshchandlerblog.blogspot.com
http://tinyurl.com/6x3gbw
Gentlemen, you are not the center of the universe.
I see more than a few people commenting here who might be feeling the same way.
I predict we will see a Scoble and Arrington backlash some time in the future, as people get tired of the games they are playing.
Frankly I am very surprised that MSFT hasn't moved already, unlike GOOG with Jaiku, they don't even have Micro-blogging in their arsenal.
Now there would be a worthy project for them to hone their skills at cloud computing and search (Twitter is badly in need of more useful NATIVE search/tag/filter/sort facilities to make the onslaught of potentially useful data, well, useful...
(What if Scoble could subsegment his follower/following lists, with e.g. "all followers who have ever used the term "branding" in a tweet", right now it's either all of Twitter with "track", or single user on Tweetscan. Also note that the overview of one's follower list is almost completely useless right now, since there is no way of sorting them in any predictable way, e.g. last-in, # of followers, etc. etc.)
Alas, since MSFT still doesn't get the Internet much less Web2.0, it's more likely that Google will move eventually after seeing that Jaiku has already missed the boat as far as mind-share/branding/positioning is concerned. Twitter(ing) has already become "the verb" for micro-blogging. Would be same as the Google Video vs. YouTube story, except that Twitter can still be had for, what, $100M or less (given their recent problems?).
Twitter already integrates with GTalk, it really seems like a no-brainer.
Or... possibly... can you say "bidding war"?!?
The expectation should be: you get what you pay for. Gmail and Hotmail have been down for lengthy periods of time in the past. And surprising as it may be; miraculously, civilization continued to progress.
Considering this, it's hardly surprising that Twitter 'blames' its 'users'. It puts me in mind of the early days of broadband ISPs imposing bandwidth caps because of a minority who abused the system with torrent downloads etcetera. It's not fair when companies do something like that, but it is at least understandable, given how this minority of people are essentially using the system in way it wasn't designed for, and so pushing it to untenable limits.
give me a break.
Dave Winer, father of RSS says “Twitter, as it was conceived, was never meant to live.”
“It’s very possible with better engineering its architecture might have gone on for a few more years, but eventually it would have hit this wall, where there were too many people posting too many twits to too many followers. The scale of the system as conceived rises exponentially.”
So is the end of Twitter getting near? I hope not. Twitter I hope that you are listening and you better start taking things more seriously.
-----------------------------------------------
Here's my two cents.
For instance there are about 100m users of yahoo messenger and usually 2-3 of them talk at a time that means scalability of 300m conversations. On the other hand with 100m twitter users who usually send messages to 100-10,000 other users the scalability required is 10,000m to 10^6m I have never known any current architecture based on webservers to handle such a scale. So according to me Twitter was never meant to live. It is like a concept car that will never see production. Users of twitter don't understand this and they don't care.
They don't know whats happening when the website is down. The sad part is that the best analysts claim that Twitter is a billion dollar company in one year of operations. There is an old saying before the days of when people understood permutation combinations. One peasant asked a king to give him rice equal to the total amount gotten by placing double the number of rice grains on a chess square than the previous square, starting with one rice grain. There are 8x8=64 squares. We seriously need to visit grade 7 mathematics.
I know of only one News/Messaging system that supports around 1 billion users sending messages to all 1 billion users each. Thats a scalability of 10^12m. It is not Web based but rather on a massively scalable serverless P2P architecture based. The team is soft spoken and when I last talked to them I was told that they don't care about money or hype or fame but rather for just the passion of next generation global systems that will stand the test of worldwide use. Its called Mermaid News Mermaid
They have other softwares too but this post is about Twitter and Messaging. Once everyone comprehends basic mathematics that goes behind scalable algorithms they would go past the flashy screen and hype to actually want a system they can trust. To the analysts I would say it is easy to create a business plan, create a hype and raise $20m funding it is far more difficult to create something of use.