Ask HN: Do you know any useful "Back of the Envelope Calculations"?
8 comments
One I use often is that 1 Mbps of IP packets ~~ 10 GB / day of data. (1 Mbps is 10.8 GB/day, but TCP/IP overhead will typically eat up at least 0.4 GB and usually close to 0.8 GB from that.)
Another, more often financial calculation: If you have X% growth, it takes 72/X time periods to double. (Accurate to within 1 time period for X% > 2%, and accurate to within 10% for X% < 29%.)
Another, more often financial calculation: If you have X% growth, it takes 72/X time periods to double. (Accurate to within 1 time period for X% > 2%, and accurate to within 10% for X% < 29%.)
Using 70/X will also work just fine.
Yes, and for small values of X, 70/X is more accurate, too. 72 has the advantage of being a multiple of 2, 3, 4, 6, 8, 9, and 12, whereas 70 is a multiple of 2, 5, 7, 10, and 14 -- knowing both of these gives you a whole-number answer for any growth rate under 15% except 11% (correct answer is 6.64) and 13% (correct answer is 5.67).
Memorize powers of 2 up to 9
0 1 2 3 4 5 6 7 8 9
1 2 4 8 16 32 64 128 256 512
2^10 = 1K, 2^20 = 1M, 2^30 = 1G etc.
ex. 2^23 = 2^20 * 2^3 = 8M
ex. 2^37 = 2^30 * 2^7 = 128G
ex. 400K: 256K < 400K < 512K <=> 2^18 < 400K < 2^19
It's trivial but useful that an hourly rate multiplied by 2000 gives a year's worth (for a standard working year).
Yep, that one's good, subject to the caveat that the 2000 is really 1800 in Europe (more holidays) and 1500 in France (lots of holidays AND a less hours per week).
Related and very useful: Consulting rates should start at annual income / 1000.
Related and very useful: Consulting rates should start at annual income / 1000.
Not quite a mental calculation, but still in the spirit: One rough estimation I often find myself using is that a US dollar bill is six inches long. Using either folding or flipping, you can approximate most distances in the wave-your-arms-about scale. Be it thinking about a new physical device or ball-parking blinds at Home Depot, I use this all the time.
I use revenue per employee as a rough guide to the type of company in question (is it human-based or tech/marketing-based? is it a startup or an established business?). It's simple to calculate, and revenue is usually less sensitive information than net income per employee (so easier to get for non-public companies). My feeling is that for the average (non-public, not particularly successful but OK, non-automated) company the average salary is about 1/3 of that (more or less). Actual net income per employee of course depends on the degree of automation and success.
Example tables:
http://www.jbryanscott.com/2009/02/07/nasdaq-100-revenue-per...
http://37signals.com/svn/posts/2283-ranking-tech-companies-b...
Example tables:
http://www.jbryanscott.com/2009/02/07/nasdaq-100-revenue-per...
http://37signals.com/svn/posts/2283-ranking-tech-companies-b...
I've always liked this page - http://www.vendian.org/envelope/
and the dots clock is interesting as well
and the dots clock is interesting as well
The rule of 72 for me is a classic that comes to my mind.
http://en.wikipedia.org/wiki/Rule_of_72
http://en.wikipedia.org/wiki/Rule_of_72
Two:
* Golden ratio is also miles to ks. Which means you can use Fibonacci to convert miles to ks.
* Divide by square route of two for compound doubling time.
* Golden ratio is also miles to ks. Which means you can use Fibonacci to convert miles to ks.
* Divide by square route of two for compound doubling time.
Thanks