Jedberg, I'm sorry I didn't respond, but I don't have a record of getting an email. Corrections are always welcome. A lot of articles are summaries from multiple sources, so they aren't made up (I'm not sure how I would even do that), and the sources are always referenced somewhere. But of course I could screw up. And I don't make an effort to contact people anymore because seldom is there a contact point and they rarely respond or follow up. Which is why I rely so heavily on existing materials like blog posts, presentations, slide decks, forum threads, and whatever else I can find.
I tried to contact you through the form on your site. Regardless, I wasn't accusing you of making things up, I was just stating that sometimes you don't use very reliable sources (like blog posts that may also be unreliable).
I don't work at reddit anymore, so I can't speak to their setup, but if you ever need help with an article on Netflix, let me know!
Yes the a tilde diacritic is missing but its close enough. I also remember reading die in German sentences at the start and getting very confused in my brain for a while. Maybe its just me. :)
I interviewed with them in their Vancouver offices. They are near the top of the Harbour Centre downtown, and have a crazy beautiful view of North Vancouver and the mountains.
I would say there were about 15 people in the office at the time (1.5 years ago)
Markus is a slightly weird Rain Man kinda guy, but is obviously very smart. A bit smug, though likely deservedly.
He has a hot girlfriend, that is the head of marketing or something.
It was a three on one interview, which isn't very welcoming. No whiteboard, no code writing. Kinda glad I didn't get it, actually.
"POF has one single employee: the founder and CEO Markus Frind. Makes up to $10 million a year on Google ads working only two hours a day. 30+ Million Hits a Day (500 - 600 pages per second)."
I find this more interesting than the parent article; although it seems clear hiring more people helped them expand (and handle that expansion) a lot, a single employee/founder doing that much is crazy impressive.
No built in components from ASP are used. Everything is written from scratch. Nothing is more complex than a simple if then and for loops.
and...
In the process of getting rid of ASP.NET repeaters and instead uses the append string thing or response.write. If you are doing over a million page views a day just write out the code to spit it out to the screen.
I think this goes a long way to explaining how this is all running on IIS on a seemingly "small" hardware infrastructure. Last company I worked for reimplemented a "classic" ASP app which was mostly in-page SQL statements calling stored procedures, with Response.Writes building the HTML. The new implementation, in ASP.NET, used .NET components, an ORM, an elegantly abstracted object model, agile methodologies, etc. It was slower than sludge compared to the old app.
You can get too carried away with abstractions in .NET. Where I work it was the norm to have multiple layers, abstractions, frameworks etc... A consequence of this was 30 minute build times etc... I think the people working on the stuff initially just wanted to play with every language feature under the sun.
I ditched it all for a few simple layers (mostly MVC with LINQ to SQL) and as a result the application takes 2 seconds to compile, is very fast at runtime, and has no magic going and is easy enough you could hand it over to a junior developer to maintain.
No, they have a jobs page (http://www.pof.com/careers/) and it lists director of technology. Another user posted above an inc article saying that he had 4 employees in 2009 and planned for 30, so maybe he has 30 now!
Nobody's saying it's impossible — just that like many other possible sources of revenue, such as lottery tickets, it isn't very dependable. If you're able to get 200 million pageviews a day, all sorts of avenues for making money are open to you. For most prospective businesses, though, "I will get 200 million pageviews a day and be able to maintain it practically by myself" is not a very safe assumption to make.
Not sure your point, I make my income from advertising (well company does and pays me, same thing right?) my fascination was with the single employee part.
So around 2.3k HPS monthly average. Assuming an average peak is 10 times that, 23k HPS seems a little little high but doable for a handful of servers (after load balancing and caching come into play).
One note says "everything is dynamic, nothing is static" which sounds weird if they're using Akamai for anything but images. Images are claimed to be served at 100K HPS, which would be difficult for a handful of servers, especially with the dynamic content on top of it. It comes down to caching and hardware as usual.
It isn't to say a lot of people thought it was impossible, but it is just very unheard of. Anecdotally, most of the people I know hadn't heard of it either.
Slashdot, Facebook, Wikipedia, etc were the very large sites talking about what they used over the years and it was all about the open source stuff like Apache, PHP, etc.
Then I found out, on HN, that POF used IIS and it was mind blowing. You would think even Microsoft would promote its successful uses more like that. Until POF, I too had assumed IIS was only for 1,000 to 50,000 enterprisey setups. I didn't think it couldn't be used per se, but I just hadn't ever heard of it being used that way until POF either.
I don't know what enom is like anymore, but when I used enom back in 2006 to 2009 I wouldn't have used them as a poster boy for anything technology related. Sporadic API failures, unexpected "lost" registrations etc. etc. If I had to choose again, I'd be choosing OpenSRS
Not really. Contrast the cluttered mess that is PoF to OkCupid, which has decent usability and reasonable information architecture. OkC's design is also better at extracting the right information from more users more consistently, meaning the resulting corpus of data is more complete. Which makes for a better product, since it's easier to find the sort of people you'd be into.
PoF is successful in spite of its questionable design, not because of it.
Or an example of the network effect overriding other condsiderations. PoF got big before OKC did, so that's where the people are.
Even with that factor though, I found PoF borderline unusable after a little experimentation and stopped using them in favour of OKC. I can't believe I'm the only one like that, or that there aren't some who simply dropped off PoF into nothing.
If that crappy site can extract value at such a high level, I'm sure a few reasonably small tweaks could extract significantly more.
At the scale pof operates at you have to A/B test everything because some things are very counter-intuitive.
My favorite anecdote of his (tried to find it on his blog but google is failing me) was when he briefly fixed the problem I hated most about the site: the aspect ratio of the profile pic thumbnails being off.
What happened is that it cost him lots of money. And the explanation as to why makes sense too: when the aspect ratio is messed up people can't tell at a glance what the person looks like so they click through to people's profiles a lot more instead of just scanning the top bar/search page. When your business model is based on page impressions getting people to load more pages means you get more money.
So in that case at least it's better for his pocket to have something that's obviously broken (and easy to fix) on the site.
If your business model is dependent on being just awkward enough to use that you generate extra page impressions but not awkward enough that you drive away traffic, to borrow an aviation term you're flying awfully close to coffin corner. Particularly when you're dependent on network effects; if enough people start discovering something that's less awkward for them (such as OKC), the network effect will start working against you and you're suddenly so 2004 with a database of inactive users and a reputation for being awkward to use. Plus, if you're dependenet on awkward usability inherently generating more pageviews, surely by definition those pageviews are also more rapid so the time each ad is being displayed to each user is lower, so the savvy ad buyer would also be paying less.
Now, I could make the case that PoF has ridden the wave for long enough to build Markus enough money that if it all fails tomorrow he's still made for life and only really lost pride, but....
Take all this with a grain of salt, as I'm about to get vague and handwave-y. According to some article I read [citation needed], the fact that their interface is crap actually drives more pageviews. E.g., that profile pictures are distorted is searches actually encourages people to click through to profiles to see better-looking photos.
So poor usability actually is a feature.
Which comes back to the old argument about advertising-driven revenue models resulting in poor user experiences, since the "user" from a revenue model is the advertiser and YOU'RE the product.
too bad both sites use the old stereotypical pictures+profiles approach to dating unlike flirtrs which actually makes you go meet people (cause you know, if you are trying to meet people it helps to go meet them)
Frind was making $10 million USD per year when he had 1.2 billion page views. If his revenue scales directly to pageviews (which I think it does), he should be bringing in about $50M per year.
I know Markus from good 'ol webmasterworld days when he was a active affiliate marketer. He is really a bright guy and POF's initial growth was mainly from SEO (until the network effects kicked in).
If i remember correctly until a few years ago the site ran on a single webserver and db server!.
At one point Highscalability covered the reddit architecture, but he didn't actually talk to anyone at reddit about it, and got a lot of it wrong.
I tried to contact him and get it fixed, but he never responded to me.
So I have to wonder where these numbers are coming from.