More Lies, damn lies, and statistics

Nothing is more meaningless or misleading then a statistic without context.

For instance, take this one:

Virtual Products & Goods Sales Reach $5 Billion in China vs $1 Billion in US in 2009

This article goes on to blather about micro-transactions and virtual object sales as a primary monetization model. But the link between these two ideas is never really established or demonstrated.Consider THIS statistic in light of the one above...

Over One Million Gold Farmers In China

So, how much of that 5 billion dollars above is in fact being driven not by F2P/virtual object transactions at all, but by the sale of gold coming out of traditional MMORPGs? And not actually being sold to Chinese but, in fact, to the west?

Based on the numbers above I'd hazard to guess a lot.

Emperors new clothes and suits made of suede

On numerous occasions I've heard it claimed that Free to Play (F2P) games are generating Average Returns Per User (ARPU) per month in the double digits.

I've always doubted these claims. They make no sense. Its well known that less then 5% of the users of these games pay *anything* at all. In order to return a monthly ARPU of even $10, this means that the average spend of those who DO spend money must be at least $200 a month.

How many people do YOU know that spend $200 a month, month after month, on micro-transactions? The very idea boggles the mind.

Where then did these inflated values come from? I found the answer in this article.

"http://lsvp.wordpress.com/2008/06/09/successful-mmogs-can-see-1-2-in-monthly-arpu/"

The answer to the mystery is in a footnote... almost an afterthought....

"* Note that this is based on monthly users. Many MMOGs calcuate their average revenue per user (ARPU) based on Peak Concurrent Users. On this basis, ARPU can be more than an order of magnitude higher than the $1.40 guideline."

This is nonsense. An accounting trick. The Peak Concurrent Usage (PCU) of a paid for game is only 5% to 10% of the user account population. If WOW (or most other pay to play MMORPGs) calculated their monthly ARPU this way, it would be between $150 and $300!

The PCU of free games is, logically, an even lower % of their account population as the incentive to come online is lower then if you are paying monthly for the privledge, and accounts of people who quit the game are never removed.

The F2P Emperors' new clothes, it turns out, are far less then their tailors' would have you believe. And as for suede suits, that goes to an old crude aphorism about gullible people I will chose not to repeat.

Saying goodbye to Sun

Today, Sun Microsystem officially ceased to be.

Sun was a great engineering company that was responsible for some great innovations. It will be missed by myself, and I expect many others.

Through Oracle's official announcements as well as what I am hearing from various friends in Sun, whatever gets folded into Oracle will not be the Sun we knew and loved. Frankly, I find some of Larry's choices to be a bit bizarre, but he bought the right to make bizarre decisions with his cold hard cash.

I do think there is a lot of gold being thrown out on the street in terms of both people and projects. It is my fondest hope that both get picked up by others with the vision to make them the successes they deserve to be. And in terms of what is being kept, well, I predict some indigestion in 18 to 24 months, but I could be wrong there, too.

Kenai to the rescue

I don't know if this has anything to do with the oracle Sun acquisition or not, but Sun has a new community development site up in beta and its HOT.

I never cared much for java.sun.com. I felt the Oriely software it ran on was ugly and ill fitting to my sorts of projects. But Kenai is awesome. Its a modern interface, has every feature I could imagine I'd want AND has full Netbeans integration. Best of all, its really a site for the community.

Unlike google code, who want to force you into releasing your project under licenses they like, Kenai supports every OSI license that exists. PLUS if you want to use your own license, you can. Now thats what I call community service.

Speaking of Netbeans integration, Ive gone back to NB with NB 6.8 and must say, I am VERY impressed so far. They may have leap-frogged eclipse again. The Scala plug-in needs some work still but is definitely better then the one for eclipse.

Which brings me around to the Scala work I've been doing. But thats for another post...

Conclusion: If you need free open source project hosting I'd *strongly* suggest you look at http://kenai.com

Saying Goodbye to Google Code

*sigh*

Just when I thought I had found a single provider for 90% of my net needs I could be happy with, I came across the following thread:

http://groups.google.com/group/google-code-hosting/browse_thread/thread/1714c5c0ef5d9f9f/3fc51c3c0ca497ec

Particularly this...

"Basically the answer is when I, Fitz, Greg or the team think it is

popular enough. I know you guys think we don't like it for nefarious
reasons, but what you're missing is we dislike -all- new licenses that
are unpopular. They lead to bifurcation of the open source development
world and that is a high price to pay.

I personally think the AGPL is deeply flawed, and I've commented on
that on my own blog and on others, but that really -doesn't- matter.
If the AGPL gets to be popular, like lgpl or bsd popular, than we'll
certainly offer it as an option on code.google.com, "

The other posters go on to show that AGPL is used on a great many sites and ask how many it takes to pass the "popular" test ... and "Chris" just stops answering.

Now GoogleCode is supposed to be a community site, but here is a google-person stating that a license has to meet *his* particular standards or they won't allow community members to use it

Now I don't know who Chris is, For all I know, he could be a Google founder, but I don't really care. The point of community is not to arm-twist everyone else into doing what you individually want. The idea that google finds such an attitude acceptable has to make me seriously rethink just how google-dependant I really want to become.

As for my AGPL project, I guess I'm headed back to SourceForge who use the yardstick that, if its recognized by OSI, its an okay license. Frankly, I've never been thrilled with OSI either, but its a better yardstick then Google's.

I am Joe, Dragonrider of Deathworld

So, we went and saw Avatar last night at the 3D IMAX.

It is a beautiful movie. It absolutely sells the mix of 3D animation and live action on a level I haven't seen before. The art direction is simply stunning and highly imaginative. The 3D is used not to poke you in the eye but to immerse you in the space. It was well worth the few extra bucks we paid to see it in large format 3D. It is, in a word, breathtaking.

The story is well written and executed but is mostly a mix of semi-standard science fiction plot vehicles. I wont go into spoilers here beyond the title of this blog, but suffice it to say there are no ideas here that will excite or surprise anyone with a decent knowledge of classic speculative fiction.

I wouldn't call it a great movie, in the way, say, The Godfather is a great movie. This is, in the end, B movie fare. But, of course, thats what Cameron has always been good at. It does have stupendous A-plus production values and is worth watching for the visual spectacle alone. That the story is decent and competently told, is an added bonus.

A tale of three android ad networks

I wrote a simple dice-calculator app for the Motorola Droid and have been experimenting with various advertisers on the ad-supported version. Here's my capsule review of three; AdMob, Quattro and Greystripe.

Its important to knwo up front that NONE of these so called 'advertising networks' actually tell you how much they pay or for what. Therefor, you have to experiment and intuit. These are the

results of my experiments to date.

AdMob.

The Good:

AdMob is the 800 pound gorilla in this space having been bought by google. Their sdk is the most mature and easiest to integrate. Their ads are also the least intrusive. They will also run your "house ads" for your own products in your other apps for free. Finally, their reporting tools are *excellent* and really give you a good feel of the trends of usage of your app.

The Bad:

Unfortunately, they also pay the worst of all three. Although they don't publish their exact formula (none of them do) it semes to be *heavily* if not entirely base on click through. I've served almost 7,000 impressions, but only had 14 click throughs. Total revenue: $0.88

Quattro.

The Good:

Quattros ads are still fairly unobtrusive, about twice as tall as an AdMob ad. While the SDK is a bit less mature, it is a bit more automatic. AdMOb's view has to be told by your app when to get a new ad. By contrast you just place the Quattro view and tell it how often you want to change ads and it does the rest.

Quattro pays for impressions as well as click throughs and it shows in your revenues. With about 6500 impressions and only 6 clicks I already have made $1.44. Quattro also encourages a faster ad-flip rate and thus Ive gotten those 6500 impressions in something like a third the time it took for AdMob, who I integrated earlier.

The Bad:

Quattros analytics pages frankly suck. You cna ask for number between any two dates but what it shows you is the aggregate total over those days, not numbers by day. This makes trend analysis very hard. They also don't have "house ads". Finally, while their view is easier to integrate, you also have a bit less control over it.

Greystripe

The Good:

Greystripe is different from the other two who are primarily banner-ad suppliers. Greystripe specializes in interstatials, ads that take over the entire screen for their run. They imply on their website that they pay per impression. In fact, their reporting tools don't even show click-throughs. This means you get paid for displaying the ad, whether the viewer clicks through or not.

I have found their tech support responsive, which is important given their issues. (See below.)

The Bad:

Interstitials are by far the most intrusive kind of ads there are. Therefor you won't be generating high impression rates per user if you don't want to drive your users away. I have my app configured to display one on launch and then one approximately every 5 minutes. For games, it makes more sense to integrate the interstitials between levels since you cant pull the screen away from the user mid-level.

Their Android SDK is in beta and is by far the hardest to integrate. To begin with, its not perfect yet. My interstitials don't display properly the text on the buttons to go back to the app or to the linked advertiser web-page. (I've reported this and they say they are working on it.)

The biggest problem however is that they chose to implement their ads not as an Android view, like the others above, but as a separate Android activity. This causes your app to go through life cycle call backs that are exactly the same as when a user navigates away from your app and then returns. This is simply a nightmare to deal with, impacts app stability, and needs to be fixed right away.

Their reporting tools are neither as good as AdMob nor as wretched as Quattros.

The OTHER issue I have right now is that they *say* they pay per impression but their reporting tools aren't showing it. I have 158 impressions so far and its still reporting $0 income. This makes it hard for me to get a grip on what, if anything, they are actually paying me for my efforts. Given that the impression rate is so much slower with interstitials, this is an important consideration.

Conclusions:

Right now, for income (which is why we do this, right?) Quattro is the big winner. They aren't terribly objectionable and pay quite well. If their reporting tools were better and they had the house ad option, they'd be the only banner service I'd use. As is however those values are big enough to make me keep AdMob too. At least for now.

AdMob is the second place producer. They are less then half as lucrative from a payment point of view, but their tools are excellent and have really helped me learn user trends from this first product.

Greystripe has promise, but they need a lot of work. The SDK is definitely not done and need some serious re-architecting. And I'm very concerned that their website has yet to tell me I've made any money at all.

In web-advertising, interstitials are a big money maker and there is promise in this market, but Greystripe has work to do if they want to fill that niche.