Book Review: The Filter Bubble

I got turned onto The Filter Bubble after viewing the author's TED talk on the same subject.

The TED talk gives you main idea, and that's probably sufficient. The book dives into a lot of interesting detail, some of which isn't exactly related to the core thesis, and that's part of the problem I had with it. The idea is sound but the book is a somewhat meandering exploration of the idea... along with other things the author is interested in but that are unrelated.

The thesis is as follows: In order to better serve users, search providers, social networking sites, and other information sources are providing personalized data feeds - feeds tuned to their preferences. As these become our primary sources of information, it results in a feedback loop where we only see what we like, and what we see influences what we like. He borrows danah boyd's analogy of an all-sugar-and-fat diet (it might be what you crave, but it's not good for you), encouraging us to think about ways to eat our digital veggies.

This is not new of course. The advent of television brought about similar paranoia. However, there's no denying that it's true to some degree and the fact that it can be dialed in to each individual user makes it credible. The paranoia is seductive to give into. Even if you don't there's some interesting stuff in the book, though there are also some flaws.

Pros:

• I learned a lot about how modern internet advertising & site personalization work. I'd heard of companies like Axiom but didn't know what they do.
• The book does a good job painting a picture of some possible outcomes of personalized search and personalized advertising (e.g. think of tailored political ads, for example, and the complexities of holding them accountable to telling the truth).
• He does a good job explaining some basic concepts around programming and technology in layman's terms. Not much use to me, but I might think of recommending it more easily to a relative or non-techie friend.

Cons:

• The author delves into a lot of other areas having little to do with 'filter bubbles'.
• Those areas that do are taken too far, and consist mostly of his own 'what ifs', rather than consulting research and/or data on the subject.
• The solutions proposed are weak. Telling people they should try to consume responsibly, out-smart the personalization-bots, etc, all seem like they'll fail and/or fall of deaf ears. Suggesting maybe there could be an ombudsman or some regulation seems like a bit of a cop out without proposing how those might work.

I guess I'd say most will be better off watching the TED talk to get the basic idea, and then reading the book only if they want see how deep the rat-hole goes.

The Filter Bubble: What the Internet Is Hiding From You

Book Review: Secrets of the Moneylab

This book surprised me and is probably one of my favorites of the year, at least as it is really timely for our industry, as I'll soon explain.

Technology companies generally have laboratories in which they do research & experimentation. Game developers often do so as well. Perhaps not formally in R&D labs (though some of the larger publishers have them), but certainly doing experiments for new rendering techniques, AI models, physics experiments, etc.

For all this though, it's surprising how little science is put into the business side of the business. Things like pricing, to pick one example, are often done using some gut-feel starting point and by following competition.

Kay-Yut Chen, author of Secrets of the Moneylab, runs a lab at Hewlett Packard. He is also an economist. The lab he runs does research on behavioral economics and it's application in the tech industry. The book runs through experiments they did on consumer pricing, marketing, supply chain management, and much more. Also, they look at a number of different experiments from different industries.

Why is this so timely?

Consider the point made by Neal Young of NGMoco at his GDC2010 keynote, where he said that for the first time in the games industry, the business model is in the hands of the game designers. Add to that the fact that digital distribution channels and competing appstores will mean that developers have the opportunity to try many different experiments as they bring their games to market on different platforms (Something Dave Edery at Spry Fox has been doing).

Success on these emerging platforms is going to come from people's ability to put some science into the business side of their business, and this book provides an excellent start to getting your head around that kind of thinking.

Secrets of the Moneylab: How Behavioral Economics Can Improve Your Business

[note: I'm post-dating a couple of these posts as I didn't have a chance to sync them while I was travelling, and I like to track by when I read the book]

KZero Virtual World dataporn

Raph points us to Kzero (a market research company looking at virtual worlds) and their awesome data visualization of the growth of virtual worlds. I'm a sucker for data porn, and this shows up the worlds split by age demographic, size, and launch date.

Would love to have this as a poster, but for now, here it is in pieces (note that some of these are from last year - guess you have to pay them to get the whole, er, pie.

It's not the size of your installed base, it's how you use it

This was an interesting graph up on Joystiq, contrasting the growth of installed base between platforms.

While interesting, it's not exactly intellectually honest. For one, the iphone saw a lot of hardware refresh with the same customers upgrading to the 3G/3GS,so some of those are repeat users. Yes, it's still units sold, but for purposes of installed-base discussion, this is relevant.

For another thing, if you are going to talk "consumer tech", then you need to look at other cell phones, DVD players, etc. If you are looking at game platforms, then include the gameboy, the Windows PC. etc. Not sure any of these numbers would beat that curve, but it's worth including (though this example shows that the GBA beat the Wii's growth curve in its first 10 quarters. hmm...). Finally, the attach rate and SW ARPU would also be apple to oranges.

Still, even with all these caveats, it's an interesting chart to consider.

Dataviz musings

Short version of this blog post:

This is a pretty cool little data visualization on box office receipts over time. Not very useful, but cool and fun to play with.

Slightly-less-short version of this post:

It sure would be cool if we had a version of the above data visualization that was interactive (more than scrolling), say like this, and that we had it for, say, game releases & sales data over time. Oh, and for like 10 years worth of data. Hmmm. Yeah, that'd be nice.

Lengthier version of this post:

I've been wrestling with some gnarly data visualization stuff at work, trying to get some trends to pop out of complex data sets in a simplified and obvious way. It's been a pain, I've spent a bunch of time looking at different chart models like circular histograms, donut charts, radar charts, etc. None of them really does what I need, so I'll just have to resort to building something myself.

While looking at all these things though, I've been thinking a lot about how dramatically the impression the visualization can sway depending on the choice of how it's visualized. Also, on how much better the data can 'pop' when its interactive. The example above lets you scroll back and forth through time (do so and the holiday/summer blockbuster cyclical nature of the box office is plainly obvious). Also, mousing over individual movies lets you see their rate of decay and staying power (check out films like Forest gump or Sixth sense, for example).

I'd like more tools though. Zoom. Band-pass filter. For example, is there a macro level cycle at play? (like lemmings!). Or did the advent of VCR, DVD, Cable, Blue-ray have any material impact?

Of course, I care about having all this for games more than movies. There are a couple things holding us back.

The biggest is a lack of decent data. As I'd previously mentioned, we don't have The Numbers for games. We have NPD data which is NA only, lousy for PC, decreasing in relevance as DLC, subscriptions and digital distribution gain in relevance vs retail. Other research groups publish numbers but it's all pretty fragmented and worse, its expensive. 

The opacity of the online services sales figures also doesn't make things any easier. VGChartz does a half decent job deciphering numbers but that's only an artifact of how leaderboards work consistently across XBLA titles. Good luck doing the same with Steam and it's brethren.

The second thing we need are better tools to parse the data, especially if it's coming from numerous asymetrical sources. Things like Motioncharts (Google docs' integration of Gapminder)can help but then someone needs to go mash that data in there in a useful way. The good news on this front is that there are a bunch of dataviz apps that seem to be waterfalling down from nichy segments of the market to general business usage or integrated into the cloud apps (motioncharts being an example of exactly this).

The third thing we need is for someone to build gaming's equivalent of The Numbers. If this stuff exists in a few spreadsheets buried within MS or EA, it's not of nearly the use it can be when the whole net can dig in and start doing some archeology on the data. 

Anyhow, I've got a big spreadsheet to get back to...

Casual Connect Content online

Jessica and the gang over at Casual Connect (the spry little casual games conference that's been growing in size and depth while also touring all of Europe's red light districts) has posted a TON of content from past conferences over at their website here.

Well worth perusing if you are interested in learning about trends in this part of the games industry.

The Numbers

I was in SFO last week and hooked up for a conversation with a developer friend after my meetings, etc.

Among the many topics discussed, he raised some concerns about our industry's lack of "sharing" numbers. (Further conversation defined 'sharing' as 'widely available numbers through sharing/leaking/research/analysts/etc'). The conversation was precipitated from a discussion of Valve's disclosure of some numbers last week, the coverage of which didn't discuss some of the titles/genres that haven't fared as well on Steam (some titles that have been released on PSN, WiiWare, etc, are rumored to have fared far better than on XBLA's 

I pointed to VG Chartz as one example of how people are making some numbers available (whether released or reverse-engineered in the case of XBLA titles).

As a comparison point, he pointed me to the AWESOME site, The Numbers, which covers the movie industry and has just about every data point you'd ever want to pull.

He cited this as an example of a different attitude to sharing numbers within Hollywood. I beleived it was a symptom of supply and demand. Bigger industry, more demand for the numbers, more people figuring out ways to make money off covering that scene. I think we'll get there over time, though a change in attitude could possibly accelerate it.

Now, as a cleansing sorbet: a couple tidbits off this site:

For those that continue to beleive this crazy notion that games are bigger than hollywood (less beleived these days), I offer the following:

GTA 4 was projected to do as much as $400M. Some claim WoW as the biggest with perhaps $1B in lifetime revenue.

Titanic did $600M[corrected from B] in box office receipts. Wall-E did $112M... in DVD sales... in two weeks.

Oh, and the top grossing *franchise* of all time? James Bond at almost $5B dollars WW gross to date. Narrowly beating out Harry Potter and Star Wars. I did my bit by going to see Quantum of Solace last night.

We are clearly still a spec on Hollywood's radar. For now...

NPD's track game subs: Please don't get it wrong!

GamesIndustry.biz reports that NPD is going to track subscription data around games.

This addresses a long-unaddressed market demand for data around online games. Many have asked if the PC game business is as unhealthy as retail game sales indicates, or whether the market is in fact just changing. (Whether all of PC gaming's revenues are being sucked into a giant WoW crater and whether that is healthy for the ecosystem or not is another matter entirely).

I applaud NPD's efforts, but I do worry that they are going to mess it up. A couple of things that make this space complicated to cover:

• Game service subscriptions vs individual game subscriptions. e.g. GameTap vs World of Warcraft.
• Revenue from non-subscription sources. e.g. Advertising, In-game item sales, etc.
• Where the consumers are located. e.g. US consumers subscribing to/playing Korean games and vice versa.
• Hardcore vs Casual. I'm sure they'll cover World of Warcraft. Will they cover Pogo? Likely. Will they cover casual game subscriptions on MSN Messenger? Less likely. Club Penguin?
• Platforms. Does an Xbox Live subscription count? Is that game revenue or platform revenue?

According to their press release, they are polling "thousands of gamers ages 13 and older who are members of NPD’s online panel of 3 million consumers", which seems like a prescription to messing up a couple of the areas above (e.g. just missed most of Club Penguin's user base, most of the casual audience, etc)

Anyhow. Steps in the right direction, but still I worry.

...but we make it up in volume!

Robin points us to the "Develop 100", an interesting attempt by Evolving Media to build a list of "The World's Most Successful Game Studios" by (according to their site):

Develop 100 ranks the world's games development studios based on the revenues their products made at UK retail in 2006.

While and interesting idea, it falls short of the mark for a couple reasons:

- UK only. Meh.
- Retail only. ('We rated the most powerful transportation vehicles by the amount of hay they consume in a day!'). 'nuff said.

And of course the biggest issue (and granted, hard to put THIS list together) is that I'd much rather see the list ranked by contribution margin rather than just revenue. To borrow Raph's metaphor, (which is also his business plan, I guess) I think you'd see a lot of small mammals ranked higher among the dinosaurs.

And as Robin points out, it'd be interesting to see this ranked by a number of factors. QoL, GameRankings average, etc, etc.