Powered by glolg
Display Preferences Most Recent Entries Chatterbox Blog Links Site Statistics Category Tags About Me, Myself and Gilbert XML RSS Feed
Saturday, Oct 09, 2010 - 02:57 SGT
Posted By: Gilbert

Recaps

"This is a tough holiday for Rahm Emanuel because he's not used to saying the word 'day' after 'mother.'"
- President Obama on Mother's Day for Chief of Staff Rahm Emanuel, who recently quit his post to run for Mayor of Chicago, raising the political hopes of certain members of 4O


Well, it's been a long couple of weeks, where I again experienced the difficulty of beginning a 50 page-report and the selfsame difficulty of stopping once the page limit nears, through sheer momentum; then again, the TeX skeleton alone gives eleven (mostly empty) pages.

I also learnt that dust specks on the computer screen can get nasty in Computer Vision applications.


Google Instant

A fortnight is an eternity in the online world, with Google Instant changing much of the world's search experience more drastically than any improvement in, well... probably years. And the irony is that it's probably not even hard to replicate (i.e. it probably wouldn't have been hard to implement for, like, years? once the idea came into being).

What Google Instant does is to update a full page of search results near-instantly as the user types his request. Now, automatically suggesting terms that might complete the input in a search box isn't new - heck, my Grilled Birds site has an implementation (but only kicks in once at least four characters have been entered). The groundbreaking innovation was, then, updating an entire results page instead of just possible search terms.

As for what's really tough about that... really, nothing much. It's essentially taking the most likely term, and preemptively sending it to the Google servers and displaying the response. Technically, it's a simple concept that I daresay most competent web programmers could whip up in half a day (using either AJAX or homebrewed solutions), with the major practical concerns being the bandwidth at both ends.

But in this day and age of multiple YouTube/Bittorrent streams, most users wouldn't miss the measly dozens of kilobytes that gets sent each time they type a few more characters into the search box, so that's settled on the client end. As for the server end... Google's an 8000-pound behemoth that probably controls more network infrastructure than most nations.

So it was a very simple concept that found its time. Are there any more of these lying around?


Two Talks

It was kind of funny that I found myself at the university more during recess week than during the semester proper. I spent part of Monday attending a talk on the EndNote software, which does reference management. Yes, students and researchers don't have to manually juggle the numbers when they put citations in their papers (which, I am ashamed to admit, I did some years back). Instead, just click click click in EndNote in Microsoft Word, and it'll do all that backroom stuff for you!

Or, of course, you could just use LaTeX, perhaps with a nice environment like TeXworks, and maintain a .bib file (Note: This is coming from an uncool Microsoft mostly-supporter who will continue running with teh Evil M$oft as long as their tools work).

While EndNote has a couple of value-added uses like the ability to organize citations in libraries and groups (well, I suppose separate .bib files works too) and share citations with fellow researchers (again, .bib files can be copied...), its killer feature would be the Find Full Text option that allows a user to automatically look for and download the cited documents from subscribed online databases.

That was the only thing that could have lured me to switch from TeX, but unfortunately, NUS doesn't support the OpenURL feature. So too bad. Lala.

There isn't any reason why someone can't write a simple web app that takes a .bib file (or any reasonably well-formed list of papers) and automatically trawls Google Scholar for the PDFs (they do have a good percentage of what's available), and then maybe zips everything up nicely for download. Maybe when I get seriously bored...

The other talk, on Tuesday, was "China's Internet and Media Market - Perspectives from Five Chinese Internet Entrepreneurs". The founders/CEOs of such big (well, passably big) names as PPLive, Baixing.com and Wealink.com were invited to share their experiences and disburse tips on creating startups.

A very brief runthrough: PPLive's online TV was pushed to be ideal for longform content (i.e. traditional half-hour or longer TV programmes), as opposed to the user-generated clips (UGC) that clutter YouTube, and such content was predicted to grow much faster (and be more monetizable) than UGC for the next few years.

Also, their client software platform (i.e. a specialized program) was said to be much stickier than browser-based video services, with an average daily usage of over two hours compared to about 30 minutes.

My personal observation: If PPLive-type services are "online TV", then shouldn't be YouTube-type services be "online TiVo"? For those who haven't gotten it, TiVo is generally considered an improvement on plain old TV - say I'm a fan of a new hit series, Grey's Desperate Lost Heroes Event; why the heck would I prefer a system where I have to catch it at a fixed time, over one where I can just tune in anytime at my convenience? I must be missing something big here.

Well, there are a few possible selling points for the PPLive-type service over the YouTube-type. The user might just want to kick back and channel-surf, in which case having traditional channels would be convenient - but there's no reason why YouTube-type services can't organize clips into channels, while retaining "video on demand" functionality (wait, is that Funshion?).

The "stickiness" advantage has been covered, but that may partly be a consequence of the content on PPLive being longer to begin with. As to whether this translates into more moolah... I'm not sure.

One huge plus, for the company at least, is that PPLive uses peer-to-peer technology, i.e. they farm out part of the work of providing bandwidth to their (180 million) users (though they do maintain fifty data centres). This doesn't directly benefit the users, unless it gives them access to content that they would otherwise be unable to get.

I would say that the bottom line is: P2P makes it easier to expand exponentially without committing to expensive infrastructure, but from a user's point of view, having anytime access to their desired programming should simply be superior. Actually, PPLive seems to provide some on-demand content - as to why not all... I suppose that's a design choice of the PPLive people, and it should be good enough (especially for current event programming), until some new player finds the funding to host all their content permanently.

Finally, for online TV (and not user-generated stuff), quality content is king, so the users are only half the equation anyway. But how much will producers charge for their big-budgeted babies to be streamed? Well... but hey, they're the ones with a multi-million dollar enterprise, so who am I to question?

Baixing.com appears something like a Chinese craigslist (N.B. the self-admitted ugly text interface, instead of a drop-down menu or other widgets, was explained as the most efficient at what it does), and Wealink.com a Chinese LinkedIn, so they aren't exactly revolutionary - but then, even a tiny slice/niche of one of the world's largest markets-to-be can be a moneyspinner.

So, on to the tips: Interestingly, they independently mentioned several common points: focus, ease of use (users may be dumb, but you had better not say that) and cost-consciousness (sadly, it's no longer 1999)

More (other stuff) to come soon...



comments (0) - email - share - print - direct link
trackbacks (0) - trackback url


Next: Ten Ten Ten


Related Posts:
The Hundred Thousand
A Month Gone
Changing Pains
Seven On Seven
That's It?
Back to top

Copyright © 2006-2025 GLYS. All Rights Reserved.