Dec 07

i love wikipedia

I needed to look up something about EU membership today, and went to the Wikipedia artile on it. — and I’m reminded yet again what a wondrous thing it is. I could spend hours looking at pages like this — learning things I’d forgotten, and many I’d never known. Even the disclaimers at the top of the page are wonderful — indicating contested parts of the content, that there could be/is bias in the article, etc. Just a magnificent evolution of shared knowledge, I think, for all its faults. For me, it’s ground zero for human beings to learn again how to really read. How to read in an age of ubiquitous and cheap creation tools — how to read in an age where understanding point of view of authors is paramount.

But this article is just a beautiful example of what’s possible. In the middle of a busy work day for me, I am undone.

Dec 07

I am American (and So Can You!), by Stephen Colbert

Eh, okay. Funny in spots. Not in others. Don’t think I’d recommend purchasing it. I’m also finding that I don’t really miss the Colbert Report on TV, with the writers’ strike going on. I like his character, I’m glad he exists, but I think it’s off my nightly list of television to watch.

Dec 07

Kindle: my analysis of title availability and pricing

I was curious to see how having a Kindle might make a real difference in my life, so spent an hour this weekend looking at my 2007 Amazon purchase history, checking out which titles were available on the Kindle and which weren’t, pricing differences, etc. Doesn’t count the books I bought in airports or, you know, in real life instead of on the web.

Anyway, here’s what I found: of the 58 books I bought from Amazon, 39 (67%) are available in Kindle while 19 (33%) aren’t. For fiction, it’s 16/23 (70%), and for non-fiction it’s 23/35 (65%).

Overall, I spent $837.34 on books from Amazon in 2007 — of the books that were available on Kindle, I spent $578.14, compared to Kindle-pricing of $380.01, for a savings of $198.13. (When you include the price of the hardware at $400, means it’ll take 2 years to pay for itself.)

But here’s the kicker: of the 23,034 pages of books I purchased last year, 14,871 (64%) could have come to me as electrons instead of dead trees. Now, I’m not being naive — I don’t have the tools to do some sort of eco-analysis on the total energy footprint of the Kindle and servers compared to the relatively-more-efficient-and-developed printing industry. But I do know that there are several 500+ page books that I’m just not reading because they’re too big to drag around. Halberstam’s The Coldest Winter, for example, or Winik’s The Great Upheaval, or Follett’s World Without End. And I think with a Kindle, I would start reading them all.

[disclaimer: there’s some pricing and page count funkiness because of the timing of the analysis, availability of paperbacks now versus hardbacks then, etc. also, i should note that the real-paper catalog on amazon doesn’t seem to be the same as the kindle catalog — they’ve got lots of sync work to do there. titles were different, searching was different, etc.]

And I’ll note also that of the 58, I’ve loaned out probably a dozen this year to friends — something that’s impossible at the moment with the Kindle.

Anyway was an interesting exercise for me — I think that for someone who reads as much or more as I do, this’ll make a ton of sense. For others, I think waiting will make more sense.

Here’s an image of the spreadsheet if you’re interested…fiction at the top, non-fiction at the bottom; available titles in green, unavailable in red.

kindle analysis

Dec 07

Hybrid organizations: Wikipedia paying for illustrations

Noam Cohen has an article in today’s New York Times about a new Wikipedia initiative to pay $40 per illustration for some number of needed images for the online encyclopedia. (I haven’t checked into the story enough to confirm with the Wikipedia principals though. In the meantime, they’ve got an interesting blog about what they’re up to, fundraising-wise in particular.)

This marks the first time that Wikipedia is paying people for their contributions (although I believe they have a small paid staff working on things full-time).

I, for one, am really happy that they’re experimenting here. I don’t know (and don’t think that anyone does) what will happen, how the broader community of contributors and content consumers will feel about the change, and whether it’ll be significant for them or not.

But I am ecstatic that they’re experimenting with ways to make an already great Wikipedia even better, even when they knew there would be articles with negative overtones that would start to appear.

And I’m hopeful that there will be more of these hybrid experiments, and in particular more of them attempted at the global scale of a project like Wikipedia or Mozilla. We’re in the middle of a bunch of them here, including staffing up around the world, working with commercial partners for distribution, and generating revenue from partnerships. Not your traditional non-profit endeavor, but I think yielding results that are quite positive.

There’s a growing number of these hybrid companies operating on the Web today — we’re trying to smudge some of the distinctions between non-profit and for-profit, between project and company. They’re organizations/companies like Wikipedia, Creative Commons, the Participatory Culture Foundation, kiva.org, and Mozilla. I know there are more out there, and I’m incredibly excited that it’s happening.

One other point: while it can be uncomfortable to show up in the Times, I’m very glad that Noam and others are starting to write about efforts like Wikipedia’s and Mozilla’s. They’re not easy to categorize, really — and we’re just starting to learn how to talk about, let alone measure, the techniques, strategies and  efficacy of public benefit mission-oriented organizations who’re using Internet market mechanisms (both for revenue and for the spend side).

Nov 07

Mozilla & Firefox Market Share

This is a super-long post, so I’ll start with the punchlines:

  • We think there are at least 125,000,000 Firefox users in the world right now, give or take. That represents a doubling since Firefox 2 was released a little over a year ago, and significant growth in every country.
  • At Mozilla we view market share as an important quantitative metric that can help us ask smarter questions and build better products, but it’s only one of many
  • We have systems here that tell us approximate number of daily users, and use that information to inform much of what we do.

This is the first of many posts that I’d like to write (and see others write) about the data that we’re seeing about how people use the web. I’ll come back to some of the basic concepts frequently, as they’re building blocks for some of the ways we look at the world.

One of the most interesting things about Mozilla is figuring out how we’re doing, and where to focus and spend time. We look at many indicators; some of the most important are things like the health of addons.mozilla.org, the volume and quality of community contributions to our platform and products, and the vitality of Mozilla-related discussions in the blogosphere. 2 other more numerical indicators that we track are Firefox market share (sliced a number of different ways) and the raw number of users of Firefox.

Lots of people report market share for Firefox — it’s a succinct indicator of momentum in the market and with users, and makes a great headline to report that we’re eating IE’s share worldwide, or have exceeded 30% in Germany, or have tripled our share in China this year. And by any account, our market share everywhere is growing — extremely quickly in places like Spain and China and Brazil, somewhat more moderately in the US and Japan — but growing in every locale around the world.

But let’s go back a step: Mozilla (in all forms, including the Foundation, the Corporation, etc) is a public benefit endeavor, with the primary mission of keeping the Internet open and participatory. So why would we care about market share, something that’s traditionally associated with shall-we-say less mission-oriented motivations? In my view there are at least 3 basic reasons that we care (and probably many more):

  • It’s one indicator for how we’re doing, how people like the products that we offer, and how well we’re communicating. Not the only indicator; not the most important; but a significant indicator, nonetheless.
  • It helps us ask questions that let us support our mission better. Including these (but there are many, many more):
    • Why is our market share so high in Europe compared to the US?
      (50% higher, and more in some countries) Understanding why some countries exhibit higher usage than others can provide pointers to how to build a more relevant product for more users, and how to communicate about the product.
    • Why are so many people in China using the English version of Firefox?
      (as much as 20%) Is it just that people use English for business? Is there some perceived difference? Understanding this can help us make both our Chinese language version better and even potentially our English language version for folks using it in China.
    • Where can we use market share most effectively to help the spread of Web standards? (everywhere but Korea, for the moment)
  • We believe Firefox is a superior product, and so we want as many people as possible to use it, or at least know there’s a choice. While I’m happy that there’s increased innovation in the browser space lately, I think that Firefox has the very best blend of features, extensibility, and simplicity, not to mention the fastest security patching in the industry, by a longshot.

But market share reports have issues. The most basic issue is that there’s no way to really represent the complexity and the dynamism of the global Web — it’s just too big, with too many things changing too rapidly. Different studies have different biases: some are oriented towards early adopters (like W3Schools, who reports Firefox at 36%), some (like XitiMonitor) will bias towards European properties, and currently all studies undercount or neglect usage in Asia, South America and Africa (all admittedly tricky to characterize usage in). Don’t misunderstand: all of these studies are extremely useful and help understand what’s happening around the world. I’m just asserting that it’s very important to understand the limits of particular studies and the assumptions that are baked in.

We also look to data from individual sites to help us understand Firefox usage. Historically we’ve skewed very high on techie sites and early adopter sites — for many Silicon Valley startups Firefox use represents well over half their traffic for a significant part of their early life. Conversely, we’ll tend to see lower-than-average Firefox usage on some of the more mass market populations. It’s been hard for us to share this type of data broadly because of the idiosyncratic nature of what we see, and because most of this type of data is shared with us with the understanding that we’ll not republish. But on the whole, the data that we’re seeing shows Firefox at something like 19-20% worldwide on sites that we consider mainstream.

There’s another, subtler issue here: most of these measures are in page views, not users. (This makes sense, as by and large, the thing that folks are trying to track is market share of web properties, not the applications that you use to get to them.) As hard as it is to get an accurate read on worldwide traffic numbers for Firefox, it’s even harder to figure out the relationship between page views and users. We have some intuitions here, supported by anecdotal evidence, that Firefox users look at more pages and do more searches than typical users, but nothing that I’d actually call science.

So in addition to these indicators, we also use our own systems to get a sense of the number of actual Firefox users there are in the world (using each language version).

At present, we believe there are at least 125,000,000 Firefox users in the world, give or take.

Here’s how we get to that number. Firefox uses a system that we call AUS (Application Update Service) to keep itself up-to-date with security patches & such. Around once a day, Firefox will ping Mozilla servers to see if there’s a new update available, and if there is, it’ll present an option to users to download and install it. (AUS really deserves a longer posting — the ping is non-identifiable, respecting user privacy, and is one of the major reasons that the Firefox user base, as large as it is, is nearly all using the most recent, patched version at any given time.)

We count those pings, categorized by language version of Firefox, so we have a rough indication for any given day about how many instances of Firefox were running. (It’s decidedly rough — part of that is algorithmic, part of it is due to Firefox running in enterprises behind firewalls or other complex topologies, and part of it is usage based, depending on how people start and stop the application.) It’s rough, but it’s close, and we’ve kept track of the numbers over the past couple of years, since we released Firefox 1.5 (the first Firefox that included AUS). We call these Active Daily Users, and it’s a measure of the successful pings in a given day. Here’s a chart of Active Daily Users over the last year, since we released Firefox 2 at the end of October 2006:

Firefox AUS Since FX2

What that says is that averaged over a 7 day period, we’ve gone from 23M Active Daily Users in October 2006 to 42M a couple of weeks ago. (And we actually hit 48.8M yesterday, an all-time high that shows continued strong growth over the last few months.)

To get from the ADU number to our whole worldwide number of users, measured in terms of uniques in a given month, like most every web site does, we multiply ADU by 3. So for a couple of weeks ago, with 42M ADUs, we compute that we have something in excess of 126M unique monthly users.

This is a conservative multiplier (we think it could be more like 3.5) that we’ve gotten to by doing some of our own experiments, piecing together data we’ve received from sites who have done their own calculations, and then really testing them against the best common sense top-down tests we can. Here’s one: take our estimate of monthly users (126M) and divide by the whole number of Internet users in the world (Internet World Stats puts the current at 1.2B)  — you end up with about 10.5%, which is lower than most reports of our global traffic share. You can do a bunch of mental gymnastics from there to account for higher-page-views-than-average for typical Firefox users, an overcounting of usage worldwide, etc — but for our purposes here at Mozilla, we use a multiplier of 3 to approximate the number of worldwide Firefox users.

But, really, the absolute number of users around the world is less important than the trending. Are more people finding their way to Firefox? (yes!) How about in Brazil? (yes again) What happens over holidays? (people don’t use their computers as much — yay!) What happens when we do security updates? (usage goes down as anti-virus software updates get propagated, but recovers over the following week or 2)

Phew. That was a longer post than I thought it would be. More to come on topics like the annual autumn surge in usage, huge growth in China and Spain this year, and others.