Re: The 10**21 Problem (Part 4)

[BrowserUk]

Apart from that, for me personally, solving this problem was a journey, forcing me out of my comfort zone, learning many interesting and new things. I also enjoy writing, telling a story. I hope you enjoyed reading this story as much as I enjoyed writing it.

I'll never be really good at golf!

I have no problem with attacking a problem from all angles; running programs for days or even weeks to find solutions; and optimisation is one my most enjoyed pastimes and something of a passion. But there has to be -- at least notionally -- a practical use for the code or the results it produces.

But that's my hang-up, and is no bad reflection upon your pastime. Millions of people spend their time knocking little balls into a field and then looking for them; and often as not have to pay exorbitant annual and per-game fees for the privileged. Others write long lists numbers in books. Yet more sit around all night in the freezing cold and/or rain, on the off chance that the clouds will clear long enough to peer at fuzzy blobs of light in the night sky. Each to their own "waste of time" :)

For me, two things come out of this latest of your series:

• All the optimisation techniques, especially those from oiskuu, are extremely informative and will be useful to me in the future.
• Your statement "know your data" is one of the most oft overlooked -- and outright ignored -- missives in our industry.

  All too often people tackle tasks involving large datasets with the mindset that the must cater for the full generic range of possibilities for that data; when often large subsets of that range either cannot, or just usually do not occur.

  And in the latter case; in the rare event that the uncatered for data does occur, it can be shown that the results would be anomalous anyway.

I enjoyed following along; albeit that I came late to the party. Thank you.