The funny thing about Moore’s law is how it keeps sneaking up on you. Even though we’ve all been seeing drastic reductions in cost and increases in power in computers for years, I still find I get amazed by it.

When Moore first speculated that the number of transistors per chip would double about every 12 months he thought it might hold for 10 years or so. He later revised his estimate to 24 months, and it turns out over 40 years that that’s about right. Which, if you do the math, would be great if you could get your investment portfolio to do the same.

And although Moore was talking about integrated circuits, his law has been roughly applied (rightly or wrongly) to the inevitable increase in power and reduction in cost of all things computery.

Many things have actually made this doubling every two years seem a bit slow.

Look at the price of memory. I just picked up a 4 Gigabyte USB drive for about $15.

According to the computer science history website, the big news in 1970 was that intel’s new fangled DRAM memory was cost competitive with magnetic core memory at 1 cent/bit.

At 1 cent a bit, converting to 2008 dollars, my USB key would have set me back $1.8 Billion dollars. Glad I waited for the price to come down.

What does this mean for data analysis?  For a few hundred dollars extra you can max out the ram in your laptop. You can use it to do data analysis that no one thought you could do on a PC even just a few years ago.

I have 2 Gb of ram in my laptop right now, and I’ve been messing around with managing large data sets entirely in memory and have been very pleased by the performance I’m getting. I just ordered another 2 Gb for my machine- it cost me less than $100, and I’m looking forward to seeing what I can do with it.

Of course, Moore was really talking about CPU, not memory- but that’s all good news too- muli-core processors are just standing by ready to put all that cheap memory to use.