Debate: What’s the Reason For MySpace’s Decline?

The points mentioned are probably all valid: technology platform, location, leadership, culture.

It is interesting to compare them to the competition. Recently I've tried to answer a related question: Why is Facebook successful?

www.limitedwipsociety.ch/en/case-study.html

Facebook really is superior in every respect.

Something that doesn't get mentioned much is that the release process and operations at MySpace slowed everything down.

Facebook is obsessed with moving fast. As far as I know, they release back-end changes weekly, UI changes up to several times a day. As Jeff Sutherland once said, if you can move this fast and the competition can't, it's basically "game over."

Sorry, the technology stack excuse simply doesn't cut it.

The technology got them to 150 million users. The most obvious point here is they don't need to scale/innovate more than that, because fb is trouncing them ;)
Seriously, as pointed out in the article, did twitter fail with RoR? Did facebook fail with, what, perl!? And e-bay used to be perl/C++ before they switched to Java.

But I can see the $$$$ consultant sitting there saying, just switch to technology stack x (Java, LAMP, Haskell, Erlang) and everything will be great. Because that's what consultants do. This seems to be the Nokia solution, by the way. In other words - punt. And what usually happens is the same people write the same architecture, and make the same mistakes, or even worse mistakes, in the new architecture/stack, because they don't understand it as well.

It's a lot harder to figure out the real issue, and that is primarily that myspace does not have a Mark Zuckerberg. He's the driving force. Look at Microsoft now that Gates left, the wind kind of left their sails. The same would happen to facebook.
Now where companies do trip over themselves is poor architectural decisions, look at Digg's troubles, Twitter's, and then some unnamed company mentioned by one of the architects at Skype on a talk on InfoQ, where one of the leaders of this company made some poor decisions, causing many of the best minds to leave in droves.
So sure, there are probably some limitations in the architecture (note I didn't see technology stack) -- so fix it. From the article, it sounds very hacked together, very typical. Again, your $$$ consultant, knowing enough Java or whatever to be dangerous, says this is garbage, let's switch to EJBs, Spring, NoSQL, Javascript - and then their troubles really begin.

Now the idea Scoble mentions about LA being spread out is very interesting, and perhaps has some merit. Everybody is trying to figure out the secret sauce of silicon valley. It's still spread out, still has a transportation problems, but certainly less so than LA. But then why does LA create great movies? Wouldn't the same argument apply? Therefore, while Scoble has an interesting point about LA, he's wrong about the technology stack. I tend to agree with the others mentioned in the article, especially Seely, an insider who saw the implosion himself.

So Facebook is winning because they use PHP?

Because they DON'T USE PHP at backend.

Yes, at least once you reach a certain scale, having built your application on an open source stack is a competitive advantage.

There is the obvious advantage that open source software is easier to adapt and tweek (www.quora.com/Facebook-Engineering/What-have-be...).

A less obvious advantage is that for the kind of people who are capable of building Facebook and Google, working with an open source stack tends to be far more attractive.

Adam D'Angelo: "... most of the best engineers these days are used to open source stuff." (www.quora.com/Why-did-Quora-choose-Python-for-i...)

This is a variant of the "Python Paradox" (www.paulgraham.com/pypar.html).

scoble discusses some staffing problems related to that area and .NET skills, but merely mentions his opinion that the infrastructure was not set up to scale. whatever that's supposed to mean.

are we seriously having a ".NET does not scale" discussion based on that unclear statement here?

Without Facebook, MySpace would probably have been the dominant player in this space.

Facebook simply outsmarted MySpace.

Architecturally, Facebook came up with the idea of a social site with applications.

Because they DON'T USE PHP at backend.

Yes, at least once you reach a certain scale, having built your application on an open source stack is a competitive advantage.

There is the obvious advantage that open source software is easier to adapt and tweek (www.quora.com/Facebook-Engineering/What-have-be...).

A less obvious advantage is that for the kind of people who are capable of building Facebook and Google, working with an open source stack tends to be far more attractive.

Adam D'Angelo: "... most of the best engineers these days are used to open source stuff." (www.quora.com/Why-did-Quora-choose-Python-for-i...)

This is a variant of the "Python Paradox" (www.paulgraham.com/pypar.html).

You couldn't be more specific, could you?

As far as I know, most of the back-end services really are written in C++ (and a few in Java). There is also a lot of infrastructure code where languages like Python are suitable.

PHP is used for the user interface. A nice side effect of using PHP for the user interface is that the code is accessible to designers, who are not likely to be hardcore C++ fans.

PHP is of course not the language top CS graduates dream of, but working on a compiler for PHP is a different story.

At any rate, it's not just the language, it's the entire stack. From the OS via the database and the web server up to the GUI, it's all open source, with the advantages and the appeal that come with that.

From the OS via the database and the web server up to the GUI, it's all open source, with the advantages and the appeal that come with that.

I don't buy into any of it from random bloggers/architects and especially Paul Grahams nonsense little essays that always theme around "all the cool, smart kids love Python and you better grab em up while they're hot".

It's one of several factors. Not everything has a simple explanation.

>> It was the execution of a plan.

What was the plan?

And by the way, Lisp (Clojure) is hot again ... :-)

And by the way, Lisp (Clojure) is hot again ... :-)

It's one of several factors. Not everything has a simple explanation.

>> It was the execution of a plan.

What was the plan?

The original plan was to build a directory for Ivy League college students. Some of the investors (and even Zuckerberg himself initially) were opposed to the idea to open Facebook to high school students and the general public. Many people feared that once Facebook was no longer exclusive it would not be cool any more and students would leave.

The vision for Facebook has substantially evolved, as has the product. The same cannot be said for MySpace.

I do humbly suggest that you have a look at the case study:

www.limitedwipsociety.ch/en/case-study.html

Among other things, you may recognize that the guy I quoted on Python is not a "random blogger/architect." He was involved with Facebook since its inception and later served as Facebook's VP of Engineering and CTO in a crucial phase of its development.

Here is another quote by one of Facebook's former Directors of Engineering:

www.quora.com/Why-specifically-did-MySpace-fall...

Reliance on closed-source technology stack. This is not normally a problem in most companies, but in companies where the technical operations are world-class in size and scale, it becomes necessary to be able to directly develop and extend the technologies being used since the scale of the operation means that new technological ground is constantly being broken. Closed-source OTS technology (even with direct on-site assistance from the vendor) places the company at the mercy of the vendor, who implicitly lacks as strong a motivation to solve key scalability challenges because it is not their core business (it's just another vendor, albeit an important one). The vendor may also lack the ability to extend their technology to the scale at which it is being used, and will resist attempts to evaluate whether their technology should be replaced or re-written.

Hi Jens,

I've briefly read your case study and it is interesting, but you have to accept the cynicism :) The simple truth is we don't know. Just like the stock market we cannot predict the future. We can only try and make sense of events in retrospect, and as the saying goes everyone loves a winner :)

Consultants do this all the time. Find a success story and atribute the success to whatever it is they are disposed to that particular month. The truth is a little more complex. Market success is an outcome which emerges from a complex adaptive system (the market). Such systems are non linear with a large number of independent agents and are extremely sensitive to small changes (the butterfly effect). So no one can predict in advance which companies will succeed in the market and which will fail. This is why VC's make several bets.

For me the most interesting thing is that myspace were extremely successful at one stage, so they must have done something right once upon a time, or perhaps they just got lucky :) If Facebook were to flop 5 years from now would that be an argument against Scrum/Agile/Python/etc? Of course not!

Hi Paul,

Retrospective coherence is an interesting concept and I had actually thought about including it in the case study.

Simplistic explanations in retrospect should be avoided. I do, however, think that there are more random factors affecting the stock market than the success of Facebook! :-)

I should think that you in particular appreciate the main points of the case study, the importance of the human element (in particular Mark Zuckerberg) and the role of fast learning, which, to some extent, means building on retrospective coherence.

Mark Zuckerberg had a track record of building applications with a viral appeal, even though only Facebook made it big. Both Facebook and MySpace became popular very fast, based on their initial concept, but Facebook kept evolving and MySpace didn't.

Hi Jens,

yes. I'm not talking about randomness. I am not saying that your conjecture is unreasonable. I am saying we don't know (for sure). If Facebook is replaced by something else in the next 5 years (exactly what happened to Myspace) what then?

There are a lot of factors involved, and singling out our pet factors and suggesting that they are the key contributing success factors says more about our particular bias then it does about objective truth. We have no evidence upon wich to make such a claim, irrespective of how attractive it may be. Correlation is not proof. We are making sense of events in retrospect.

Regards,

Paul.

The original plan was to build a directory for Ivy League college students. Some of the investors (and even Zuckerberg himself initially) were opposed to the idea to open Facebook to high school students and the general public. Many people feared that once Facebook was no longer exclusive it would not be cool any more and students would leave.

The vision for Facebook has substantially evolved, as has the product. The same cannot be said for MySpace.

I do humbly suggest that you have a look at the case study:

www.limitedwipsociety.ch/en/case-study.html

Among other things, you may recognize that the guy I quoted on Python is not a "random blogger/architect." He was involved with Facebook since its inception and later served as Facebook's VP of Engineering and CTO in a crucial phase of its development.

Here is another quote by one of Facebook's former Directors of Engineering:

www.quora.com/Why-specifically-did-MySpace-fall...

Interesting article, thanks for posting.

November 14, 2007 - The feature is temporarily exposed to users due to a bug.

Dan,

Ok, now the quote gets to something a bit more interesting, but barely. I'll assume that a big part of what he's referring to his HipHop (Facebook's PHP to C++ compiler).

Hi Robert,

I liked that line, about their development process when adding the "like" button. And I would hesitate to say - that's not a bug, it's their development process! In other words, these large, but very nimble companies, like Google and Facebook, kick out certain software that is not quite ready for prime time.

my hypothesis would be that the "root cause" for fb beating MySpace, open source would be far down the line in importance

Hi Paul,

There are a lot of factors involved, and singling out our pet factors and suggesting that they are the key contributing success factors says more about our particular bias then it does about objective truth.

Hi Jens,

Agreed. Thanks for the study. My point is that a good dose of humility is always a good thing. Cheers Paul.

Jens,

First, thanks for pointing me to the presentations. They were very interesting of course. And I will concede one point - I will concede that I think you are right about the point about open source as being important.

Although I wasn't really saying it wasn't important, just not as important as some other things (as I will discuss in the last paragraph). And trust me, I am the one running around in a panic, trying to determine which wax to use before race day, like everyone else. Because faster skiers will still blow by me, but every little bit will make a difference.

I think an even more recent presentation (Sobel) also reinforces the idea of the importance of open source. But just to throw a wrench in the works, I see MySpace is using IBATIS.NET.

I think I was originally thinking this was Microsoft vs. PHP, but now I see it more as a open source vs. closed source, so that helps me understand your point more. I'm not a Microsoft guy by the way, I'm actually an open source proponent for many years, i.e. CVS, Ant, and now Rails, etc.

I think Sobel points out that most people at Facebook probably would not chose PHP if they had to start over, probably python, but because it is open source, and they can write their PHP compiler (again bolstering your point) that made it possible for them to work around the problems with PHP.

But another good example helps me see what you are pointing out. A friend of mine, way back, once ran into a problem with Informix. He argued back and forth, telling these Informix folks their code was broken. Finally, he went into the assembly code, debugged the issue, and proved them wrong. Now, obviously, not everybody can do that these days, who learns machine language, even your facebook engineers might not have the background, skills or time for that. They could learn it, but if it's in open source, it's all the more accessible, just dive into the code.

Now, one other point on a related subject. In your article, you have a section on facebook's "small batch" process. I noticed in one of the links from MySpace, that they said that they would have to make a huge organizational change, in order to fix their problems. Right there, that shows you their thinking conflicts with what facebook is doing. Don't try to make a massive change, make a small change.

So swapping out Microsoft is certainly possible, but first they have to have the culture change. They've got IBatis, how about Mono? Or maybe take advantage of the .NET environment, use some other language (Python) as stepping stone (assuming they are using VB). And as for ASP, this is where the MVC comes in to play, they could keep all their server side code in VB or whatever, but use PHP :)

So - great challenge - probably more challenging than running facebook perhaps. But it has been done - look at the IBM turnaround by Gerstner, carried along by Palmisano, and of course Jobs at Apple, Starbucks (maybe)?


-RS

Agreed.
1. Applications.
2. Photo sharing.
3. Demographics. MySpace may be where the cool kids hung out and talked about music, whereas 2/3 of my entire extended family, half my colleagues, and many, many friends are on FB.
Result is that network effects have voted FB the winner.