Nov 16

Tucked away at the back of my mind - for years now - is the little stat I
use in the column below: the size of Lotus 1-2-3 when it first turned up on
computers, versus when it was discontinued. Why should software expand like
that, and faster than hardware power and storage does?

It need not. But it does. And considering that expansion offers me  some
food for thought about other expansions - like of roads, versus the
vehicles that use them.

What do you make of it? Here's my Oct 27 column about this: The many
victims of bloat,
https://www.livemint.com/opinion/columns/the-many-victims-of-bloat-11698339390331.html

I really would love to hear your thoughts.

cheers,
dilip

---


The many victims of bloat


Remember Lotus 1-2-3? Maybe not, and I don't blame you. After all, nearly
half this country wasn't born when Lotus 1-2-3 finally died. It was a
wildly popular spreadsheet programme in the 1980s and 1990s. By the time
the new millennium rolled around, it had been supplanted by Microsoft Excel.

This is not a lament for a programme I never used. Instead, consider this.
The first version of Lotus 1-2-3 appeared in 1982. Its size was 359.56
kilobytes, or kb. The last version was released in 1999. Its size was
230.82 megabytes, or mb. That is, in 17 years, Lotus 1-2-3 grew over 640
times larger.

Was it over 640 times more useful, or efficient? Hard to say - I never used
it, remember? But I suspect the answer is "no". It may have had more
features, but I really doubt its basic spreadsheeting functions
qualitatively improved by a factor of 640.

Still, that's not the point I want to make. Here was a piece of software
that inflated like a balloon in those years. What happened to hardware in
that time?

One broad-brush metric to measure that is the well-known Moore's Law. It
says that the number of transistors on a microchip doubles every two years.
Moore's Law has held for well over half a century now, though chip
designers keep thinking they may be running up against molecular limits. In
any case, the implication of the Law is that the speed and storage capacity
of computers will double - or close to it - every two years.

Apply that to the 17-year life of Lotus 1-2-3. Doubling every two years
through that time means about a 500-fold expansion. That is, in that much
time, storage capacity of computers multiplied 500 times, and similarly for
their speed. That's a huge increase. Yet one of the world's then-best-known
pieces of software, Lotus 1-2-3, outstripped that with its own expansion: a
factor of 640 versus a factor of 500.

Which, to folks who write software or build computers, isn't really a
surprise. They know that in general, software grows faster than hardware
does. There are reasons for this "software bloat". More and maybe
unnecessary features is one. But also, with cheaper and faster memory,
programmers have little incentive to care about something that all of us
computer science students paid attention to a generation ago. That's
"garbage collection", meaning careful and optimum use of computer memory.
"The increasing capacities of the computer's RAM and storage," a 2016
article in PCMag remarks, "allow programmers to be much less concerned with
conservation."

No wonder software packages like Lotus 1-2-3 or more recent ones, and the
computers they run on, rarely seem to actually "feel" faster, even though
hardware capacity and speed increases exponentially. Software simply bloats
to fill that expanded capacity. There's even Wirth's Law, an informal but
telling principle computer scientists cite to explain this: software
becomes slower more rapidly than hardware becomes faster.

If that gives you some food for thought, it may occur to you that this
phenomenon isn't restricted just to software and computers. I remember
thinking, when I got a raise at work once, "OK, finally I'll feel less like
I'm living on the edge!" Yet my expenses found ways to rise too, quickly
accounting for my increment.

And then there are roads and cars.

Not long ago, we heard from Nitin Gadkari, Minister for Road Transport and
Highways, that India's road network increased in size from 91,287km in
2013-14 to 145,240km in 2022-23. That's a 59% increase in nine years (
https://economictimes.indiatimes.com/news/india/indias-road-network-grows-59-pc-in-9-yrs-to-become-second-largest-in-world-nitin-gadkari/articleshow/101309031.cms).
There are questions to ask about this, but let's for now take it at face
value. Also, these numbers refer to highways, but there have been dramatic
increases in road networks in our towns and cities too. For example,
Wikipedia tells us that urban roads went from 252,001km in 2000-01 to
544,683km in 2020-21; and rural roads went from 1,972,016km to 4,535,511km
in the same time. Both are better than two-fold increases, in 20 years.

Impressive? Undoubtedly. Yet think of this: We have built flyovers and
high-speed corridors and a Sealink and highways and so on - all promising
faster commutes for most of us. But it still takes me a minimum of an hour
to drive from my Bombay suburb to Nariman Point. That is how long it took
to drive that distance in the early 1990s, before the Sealink, before the
flyovers. Despite promise, despite new infrastructure, why has commuting
time stayed the same in these 30 years?

The answer lies in certain other numbers that have also risen. How many
vehicles use those roads, and how fast are they multiplying? That is
actually hard to pin down precisely, because sometimes you'll see data
about "passenger vehicles", other times "four-wheelers" and the like.
Still, there's data saying the number of "registered vehicles" (defined as
all registered vehicles minus registered two wheelers) in India was 82.6
million in 2019, an 11% increase from 2018. That followed 6.7% and 5.5%
increases the previous two years (
https://www.ceicdata.com/en/indicator/india/number-of-registered-vehicles).
Take the middle of those three increases and apply it, post 2019: a steady
6.7% annual increase gives us 107 million registered vehicles on our roads
today. Those, and the subtracted two-wheelers, use our road network in 2023.

But let's compare this rise to the road network increase above. Over nine
years, a steady annual 6.7% increase will amount to a 79.26% jump - easily
outstripping the highway increase of 59% that Minister Gadkari cited. And
over 20 years, we get a 3.66-fold jump - again easily outstripping both the
urban and rural road expansions.

That is, even a relatively conservative estimate of the way vehicle counts
are rising on our roads shows that it beats the pace at which we expand our
road network. No wonder driving into town takes about as long as it did 30
years ago. Because as fast as we build new roads, they fill with vehicles
even faster.

The same lesson as with software. The lesson cities around the globe have
learned. Bloat happens. So no amount of new roads will ever bring us better
commutes for the majority.

So if we truly want better commutes for the majority, there's this: focus
on public transport.



-- 
My book with Joy Ma: "The Deoliwallahs"
Twitter: @DeathEndsFun
Death Ends Fun: http://dcubed.blogspot.com

-- 
You received this message because you are subscribed to the Google Groups 
"Dilip's essays" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to dilips-essays+unsubscr...@googlegroups.com.
To view this discussion on the web, visit 
https://groups.google.com/d/msgid/dilips-essays/CAEiMe8qMYTGDw5ypKXpHwRmHgABLN7%3DvNc4YM6Up7gQtFf3rPw%40mail.gmail.com.

Reply via email to