We are going to make the working assumption that handwritten records are more repetitive than printed records such as bank statements. Such records are said to be low entropy and the records that we get handed in during December and January in order to prepare an income tax return are often like this.
Typically we get a little rental account with twelve receipts of £500 recorded in an exercise book. Low entropy means that there is scope for invention in dealing with this. We will type in the number 500 once, and then type in * eleven times in order to copy down the 500 entry. Alternatively, we can click on the * button on an onscreen toolpad eleven times to do the same thing. To copy down a date we use Ctrl-C, and to copy down a date with an increment of one month we use Ctrl-Shift-C. Alternatively we can click on buttons on an onscreen toolpad to do the same thing. We will have the f1 key reprogrammed from last year to say “Rental income” or similar and we can just hit it twelve times to generate the narrative.
Generally handwritten records are shorter and more predictable. We cannot use optical character recognition on them, but we can use other methods to enter them into our system which have the same mentality or ideology. We want to use the best of new technology to do it quickly. There isn’t that much of a gain with our example of a rental account, but in other cases substantial gains can be made.