Sunday, May 23, 2010

Why inferences based on percentage growth might be erroneous

What conclusion/inference is most-likely to be drawn by analysts, bloggers and journalists based on the following chart, representing the year-on-year percentage growth in a country's annual GDP (or a company's annual revenue)?

Most are likely to claim that growth in GDP has slowed down over the years.

Incorrect, in my opinion. The full picture - below - says something else (click to enlarge):

It can be seen quite clearly that even though in percentage terms growth appears to have slowed down over the years, in reality the absolute growth is actually increasing each year (over previous year). Hence, one needs to be careful while drawing inferences based on percentage figures.

An analogy to explain the above is using 'motion' (in context of physics). In the above scenario, GDP represents distance, 1 year can be taken as 1 hour, GDP for a year is the speed, absolute GDP growth is an indicator of acceleration, while 'year-on-year change in absolute GDP growth' represents jerk (rate of change of acceleration). In this case, while the acceleration is positive, jerk is negative ('growth of growth of GDP' is slowing down).

PS: This thought was ongoing in my mind for many months, and I've finally written it today.

Saturday, May 22, 2010

Masquerading as someone in Blogger comments (almost)

This concerning thought came to my mind a few days back. How easy is it to disguise your true identity and masquerade as someone else on Blogger? Too easy, it seems, as far as comments given on Blogger posts are concerned.

To exemplify, what do I need to do to make an unsuspecting individual believe that a comment is from the real "Dharma" (as visible in the image below), when it is actually by me? I only need to sign-up and create a Blogger account, and copy the name and photo of the real Dharma in the newly-created account's profile (and possibly copy other information from the real Dharma's profile).

In all probability, a reader (or the post's author) will not realize that the comment is by a masquerader, unless he/she makes the effort of clicking on the name and checking the profile (something I believe cannot practically be done for all the comments on all the posts). So, if I have a friend named X, who sometimes posts comments on my blog, it's possible for a Y to comment on a post, posing as X.

Thursday, May 20, 2010

Retroactive extrapolation of Web counter on HyperBlog

On 28-Apr-10, when I deployed a hit counter on HyperBlog, I was a little disappointed about the loss of all the previous hits on this blog. Today, I've decided to use extrapolation to arrive at an approximation of the actual number of cumulative hits till date on this blog.
So, I'm updating the current hit count of this blog to 2,338. Of course, this extrapolation is based on the (seemingly fair) assumptions that hits to my blogs do not have any seasonality, and that the average number of hits per day on HyperBlog hasn't changed materially since Jul'09.

Tuesday, May 11, 2010

Questioning the benefit of traveling on Ludhiana-Delhi flights

Yesterday, I saw an advertisement in a newspaper about the launch of flights on the Delhi-Ludhiana (DEL-LUH) route (a related news story). I was both surprised and elated, until a thought struck my mind. What benefit, if any, would flying on a LUH-DEL flight provide to a passenger, compared to traveling in a fast and luxury train such as Shatabadi or Rajdhani?

I've arrived at an answer to this question by comparing the "room to room" time (i.e., the time it takes for a passenger to reach from his cabin/room in Ludhiana/Delhi to his cabin/room in Delhi/Ludhiana, respectively) for air journey and train journey.

Total "room to room" time using a LUH to DEL flight = 315 minutes, as follows:
  1. Booking ticket: 15 minutes
  2. Home/office to LUH airport: 45 minutes
  3. Check-in/movement/waiting at LUH airport: 45 minutes
  4. Flight duration: 75 minutes
  5. Landing-clearance delay (likely): 15 minutes
  6. Exiting from DEL airport/miscellaneous: 30 minutes
  7. DEL airport to home/office: 90 minutes
Total "room to room" time using a LDH to NDLS train = 400 minutes, as follows:
  1. Booking ticket: 15 minutes
  2. Home/office to LDH railway station: 20 minutes
  3. Movement/waiting at LDH railway station: 15 minutes
  4. Train journey duration: 240 minutes
  5. Exiting from NDLS railway station/miscellaneous: 20 minutes
  6. NDLS railway station to home/office: 90 minutes
Traveling by flight saves about 85 minutes (~1.5 hours). However, two additional factors need to be considered:
  1. The train ticket costs about INR 575, compared to INR 2,230 for the flight ticket (about four times as costly)
  2. The train ticket includes a full meal, whereas this Air India flight does not
Bottom line: Flying between Delhi and Ludhiana does save time, but not enough, in my opinion, to justify the extra expense, especially when one can avoid wastage of time during the train journey by working on a laptop/reading a book/etc.

Tuesday, May 4, 2010

Internet Explorer vs. other browsers is the real browser war

In my opinion, the current browser war isn't really about IE vs. Firefox vs. Chrome vs. Safari vs. Opera. It's really about IE vs. the rest.

Why? Because even though these other browsers come from different makers, from the standpoints of intention and technology, they're all in the same camp - the camp that genuinely wants to promote modern Web standards/technologies, privacy, speed, security, ease-of-use, reliability/stability, consumer-choice, freedom (1, 2), interoperability, crowdsourcing (1, 2), innovation, openness/transparency, and extensibility (with the notable exception of Apple, whose intentions are clearly unclear).

This is as opposed to the other camp (Microsoft), which - in my opinion - probably doesn't have a strategic interest in promoting or implementing some of these principles. This camp is seen - rightly so, in my opinion - as the chief bottleneck that's "holding back the Web's full potential" (at least until IE9 is launched).

The alternative/modern browsers are certainly competing with each other (Chrome might be the biggest threat to Firefox; Firefox might be the most important competitor for Opera), but at a macro level, these mutual struggles are collectively leading to frequent and far-reaching improvements in each of the alternative/modern browsers, resulting in a steady erosion of Internet Explorer's market share.

An implication of the above definition of browser war? Once IE's market share dips below 50%, it'll suddenly become important - both statistically and symbolically - for developers/webmasters to optimize their Web applications and websites to take advantage of the increased speed and other features offered by alternative/modern browsers - even if this leads to a poor user experience for IE6/7/8 users - since the alternative/modern browsers would collectively account for the majority share of users now. And even though none of the alternative/modern browsers would individually have a majority share, they're all the same from both philosophical and technological standpoints, making it safe to treat the collective market share as if it resulted from a single entity.

By optimizing Web applications and websites for modern browsers - without caring (much) for the now-minor and ever-decreasing share of slower browsers - developers/webmasters will be doing good to more people, than they would be doing harm to.