Greek Complexity

Twitter Rant

The FT has an interesting article on Tencent, the Chinese Internet company that has grown leaps and bounds over the past few years to now become one of China's largest and most valuable Internet players. Quite well written, I thought, covering all the major aspects of Tencent's business and the reasons for its success so far.

What irritates me about the article though, as with so much press out there, is the headline: "Why Tencent could teach Twitter a few tricks". Why does this irritate me?

For starters, Twitter is mentioned all of once in the article body. The article is about Tencent, not Twitter. Why insert Twitter in the headline, Mr Sub-Editor, when the journalist herself has focused on Tencent?

But more importantly, where is the sense of proportion? Tencent is expected to have revenues of $1.5B-2B for 2009, it has 500M active accounts for its QQ messaging platform alone, its market capitalisation is $39B, it had a net margin of 39% as of 2008, and the list of achievements goes on.

And Twitter has... hype. The company itself doesn't seem to release user stats but estimates I've seen are piffly. Take this post for example which estimates that US users will number 18 million as of the end of 2009. Globally, the service may have anywhere from 35 million to 45 million users. Monthly unique visitors number in the 6 million to 10 million range. But, and here's the rub,

• "Users” are defined as people who access Twitter at least once per month. Once per month?!
• How many of these "users" are robots? No one knows.
• Twitter also has a high abandonment rate, estimated at 60% in the first month alone as of April 2009, meaning that 60% of users abandon the service within the first month of signing up! Who knows how many more follow the crowds to the exits thereafter.
• And most importantly, Twitter doesn't make money!

The only thing Twitter has done well is market itself to the media (not users). Ok, they've had a bit of a role in helping to disseminate news during occasional events such as the Nov 2008 Bombay terrorist attacks but this doesn't in any way make the company a broad-based success.

So can the media please stop with the Twitter hype? There are many many more deserving companies out there, especially outside the US.

Sixth Sense

This is by far, the most innovative interface I've ever seen and am likely to see for the next few years at least. And yet, it's conceptually so simple (execution is everything) that it's really not particularly expensive, far cheaper than what it cost them to put together those simulated scenes in The Minority Report. Plus it could be extended to several several applications (yes, that's two severals) using Internet-based plumbing that already exists today.

View the video and be wowed. And I thought the bit where he walks up to his friend was just hilarious.

Video from here.

And the Facebook debate continues

I read the title of one of Mike Arrington's recent posts and thought he was being sarcastic. Glory be, he's serious!

Mike relates user complaints about Facebook's interface changes to design by committee. That's wrong. Design by committee is before the fact, user feedback is after. I don't think any user is saying they can do a better job at designing the Facebook interface; they're just saying what they hate about what Facebook themselves have done.

So the broad ideas, vision, back-end architecture and all that is very much Facebook's job. This is NOT what users should be doing, ie users aren't involved in core design. Users don't design either Porsches or iPhones; expert designers do. What users can in turn do (generally speaking) is provide the market reaction. Give feedback on specific hitches and glitches that affect them. What Robert Scoble calls taking "splinters out of the experience".

Someone else comments on the same page:

Ignoring your users much of the time is the path to oblivion for a company. The examples cited about Porsche and the iPhone are examples of survivor bias. There’s probably 100 or 1000 examples for each of those where the idea and/or the company bit the dust.

How true. Case studies are one thing but can we look at the universe please?

Why casual mobile surfing just doesn't work

I was reading this article about DeNA, a Japanese mobile gaming start-up that plans to target the US as its next market. One comment that the journalist made was that "Americans, partly for cost reasons, typically use personal computers, rather than cellphones, to play online games and surf the Web". This, along with the fact that I was on the bus sitting behind a university student typing away on her laptop, prompted me to think to myself what my own usage was like.

Now, I have a Nokia E65, which when I bought it in mid-2007 was billed as a "business phone" because it had Wi-Fi and an office suite for working with documents. I have never, ever, not once used the office suite. Not for want of trying, mind. When I buy something, I try to squeeze every last cent of value from it. (Yes, I'm cheap.) But all I remember is exploring the office suite once long ago, never figuring out why it was important to me and then ignoring it completely thereafter. Also, I do not own a TV. Yes, I do not own a TV and haven't watched TV in ages, unless you count turning on hotel room TVs to watch the beeb.

But this is all (mostly) an aside. What I was coming to was my opinion about why people's usage of data on cellphones hasn't picked up as much as we've all been expecting it to for so many years.

Is it speed? No, I don't think so. Phones are sufficiently fast today to permit convenient, casual surfing at least.

Is it price? Yes, perhaps, to an extent. I know I don't do much more than read a couple of news articles on the way to work, but I do have the option of using Wi-Fi on my phone when I'm at home. Which is what I've been doing a lot of especially since I don't have TV. In that situation, even though my access is free, I've found myself not perusing the web as much as I'd like. Why?

Form factor: even though the E65 is a 3G phone and therefore, in theory at least, meant for data, using it for more than a few minutes at a time is just plain painful. I don't just mean eye strain; my arm actually begins to tire holding the phone up in position constantly. Even the buttons are getting to be a pain as my fingers gently begin to cramp as I scroll down page after page.

And then there's the sheer lack of mobile-ready sites. Put yourself in the average enterprise user's shoes: he's not going to www.thisgreatnewssiteforyourmobile.com! Heck, he's not even going to news.google.com, which could be expected to have a mobile-friendly layout (I've no idea -- Google thought my E65 was a scary robot/virus/hacker so it wouldn't let me view the page); he wants to go to the beeb or The Economist and for it to just work. Which, of course, it doesn't. Instead, I find that I only follow two sites now -- the NYT and the FT, both of which have very good mobile sites (the NYT's is better). Happily, these are also newspapers I respect. But what about The Economist? No mobile version. The beeb? Ditto. I don't know how many other big-name sites just don't work on mobiles, but this has to be a factor in people's usage. Sure, CNN and IHT and others have great mobile sites, but I only found that out because I'm very stubborn and am curious enough to first Google mobile versions of popular sites on my computer, and then type in their URLs into my phone using my little 12-button keypad. How many others would just give up after trying superficially on the phone? Plenty, arguably.

And all this isn't even beginning to talk about those of us who need reading glasses.

Full house?

What would the movie business look like today if movies had only been invented a few years ago? The answer to that seemingly frivolous question comes up when I can grab half an hour to bang out my next post.

Technorati Tags: movies, hollywood, internet, media, strategy

Joost!

Does anyone have a spare Joost invitation token they can give me pleeeease?

Technorati Tags: joost, tv, internet, media, internet tv

Hedge fund clones

Goldman’s Absolute Return Tracker index is a hedge fund clone that allows investors to invest in hedge fund-like strategies without the associated high costs.

Link: Goldman sets up hedge fund clone

Goldman Sachs has become the first bank to create a hedge fund replication tool in a move that could lead to a shake-up of the $1,300bn hedge fund industry. [This is a] revolution being compared with the arrival of index trackers in the mutual fund world a generation ago. Replication strategies are based on academic research that suggests hedge fund performance is largely driven by movements in underlying markets, such as equity, bond and commodity prices, rather than the intrinsic skill of managers.

So many issues come to my mind here:

• Goldman's brokerage business will no doubt face problems with hedge fund clients complaining that Goldman is biting the hand that feeds it (same for other banks that follow Goldman's lead)
• Some hedge funds will close shop when it is revealed that their strategies depend more on following prices and using leverage, than true security-picking skills
• Other hedge funds will close shop when Goldman's bargain basement fees (1%) demonstrate to customers that high fees are not a prerequisite for investing for hedge fund-like performance
• High returns will (arguably) become harder to achieve in future as democratisation of the investment process results in more money chasing the same investment opportunities. If this happens, this will be by far the biggest effect of this development.

Technorati Tags: technology, finance, strategy, software

Genomes, AI and music

[via Ventureblog] Came across a very interesting music service today called Pandora (cool name, that!) that aims to help people discover new music that they might like. This really turns me on. I'm a real discovery junkie and love trying out new things. Music is no exception.

Enough about me and let's read about Pandora.

Pandora is a streaming radio experience in which each radio station is created based upon your own choice of song or group. So perhaps you are driving to work and the song "Underdog Victorious" by Jill Sobule is running through your head. You can go to Pandora and plug in Jill Sobule or Underdog Victorious and it will play music for you that is most like that particular song or that particular artist.

And how it does that is:

Tim [Westergren, founder of Pandora] devised a way to categorize music (working with some Stanford music professors) that he called the Music Genome Project. The Music Genome is a set of about 400 characteristics that best describe the nature of a piece of music. As Tom Conrad, Savage Beast's CTO, describes it, the Music Genome captures standard things like tempo, key, etc. but also captures more subtle things like guitar picking style or "how much cowbell" (which, of course, you can never have enough of).

This is particularly interesting to me because quite fortuitously, I was talking to a researcher in developmental biology yesterday, and for the first time ever, someone explained to me in simple terms how genes work. Extending the genomic (genetic?) metaphor further, I am interested to know how Pandora works:

• Does it simply look for songs that match containing the maximum number of matching criteria as the song I specify? Because that's not how genes work. If I have a gene that is known to be tied to a particular genetic characteristic, there's no guarantee that I will actually possess that characteristic.
• Or is there some "non-linear" element to it (by which I mean, that criteria can be satisfied in ways that are not purely additive)? This, as I understand, is how most genes work.
• Does Pandora also take into account other listeners' preferences? For example, if I search for Chuck Loeb, am treated to some Astrid and rate her highly, does this affect the music served up to other listeners? This is one way of listeners passively/implicitly affecting music heard by other listeners.
• Or more explicitly, will Pandora look at listener tags as an option? At the moment, there doesn't seem to be a tagging mechanism in place. Tags could both help increase the relevance of streamed music (which, admittedly, is already pretty darn good) and offload some of the effort of analysing each song onto listeners.
• I'm a big fan of Yahoo! Launch as well -- do these two systems work in the same way?
• Most fundamentally, is this how our brains are wired? If I like Chuck Loeb's music, does that mean I will always like music that is classified as smooth jazz, has a catchy beat, has some wah-wah voices, a trumpet as lead and the occasional electric guitar solo? I'm actually asking the same question about non-linearity here, only in a different way.

Let's leave these questions to be answered by the experts. For my part, what I like best about Pandora is that it's a superb customer experience. I simply typed in Chuck Loeb's name into Pandora and immediately began getting perfectly matched music. Nothing could be simpler.

Pointer to comments from Tomi Ahonen

Tomi Ahonen has made some very interesting comments as a follow- up to my post where I talked about his predictions for mobile telecom. Just a pointer-post.

Ahonen on the future of mobile telecoms

Tomi Ahonen wrote recently about his vision of the future of mobile telecoms. Makes for interesting reading.

I contrasted this with Ray Kurzweil's essay The Law of Accelerating Returns, which I read recently. The two essays are comparable because, although Kurzweil provides much more detail, his essay essentially deals with the ways in which computational advances will change our future. So, very similar themes.

But the comparison doesn't really flatter Ahonen. His conclusions are all very interesting but what I would have been more interested to know is how he arrives at each of these ideas. After all, there is no way for us to verify any of this aside of waiting 20 years! Kurzweil does provide both empirical and theoretical bases for his ideas.

So, a better way, IMHO, for Ahonen to have presented and discussed ideas would have been to go through a logical set of arguments, with the final conclusion taking a relative backseat.

(NOTE: I'm not saying his conclusions are wrong! I'm no expert and he certainly is! All I'm cribbing about :-) is that his predictions seem to be driven by imagination, rather than reason. Not that imagination is a bad thing...)

That takes care of the more general criticism.

More specifically (and this is all shamelessly inspired by Kurzweil's essay), I don't quite agree with the thought process in some specific areas. For example, he writes:

I argued that in very rough terms and considering the performance and specifications, a mobile phone of today will be like a laptop computer 5 years ago; a desktop PC 10 years ago; a mainframe computer 15 years ago; and a supercomputer 20 years ago. With Moore's Law we can expect those trends to roughly hold true into the next 20 years. So to see roughly what kind of processing power we can expect from top-end "smart phones" of 2025, we can look at a supercomputer today - such as the IBM BlueGene/L, which as 16 Terabytes (=16,000 Gigabytes = 16 million megabytes) of memory and which runs at 70,000 Teraflops (=70 Trillion floating point operations per second) of speed.

In contrast, Kurzweil's main thesis is that technological progress is inherently non-linear. It grows not just exponentially but "double-exponentially". (Read his essay for details.) If one were to accept Kurzweil's thinking--and it is difficult not to given the amount of background he provides--then Ahonen's simple extrapolation doesn't work. The amount of progress that we achieved over the past 20 years will be much, much lower than the amount of progress we will achieve over the next 20 years. Kurzweil goes so far as to say (and "prove") that the 100 years of the 21st century will see 20,000 years of progress as measured at today's rate of progress.

I wonder what Ahonen and others think.

Link: Ottawa and 20 year future of mobile telecoms: My most challenging conference presentation ever.