The Hair Algorithm - Computerphile

Key Takeaways

we use algorithms all the time algorithms are sets of instructions to do something okay so they're just like the recipes in a cookbook they're interested for computers and two ways cuz computers actually aren't very good at making sense of things okay so they do exactly what the instruction says let's think of a really simple one washing your hair shampoo so and normally it'll say wet hair that's an algorithm you wet your hair you apply the shampoo you wash and then you rinse and each of these are instructions and each of these instructions are chained together in order to give you essentially a little bit of what we'd call a program so that makes sense you know what that means so some shampoo bottles have that so you also know what that means it means you do it twice computer wouldn't know that and would perpetually wash your hair so you'd never have dry hair so designing those algorithms and understanding them is core to how we think as computer scientists how then do you understand if things will stop what's called terminate how long will they take to do things understanding how costly those algorithms are and whether they have certain properties like they work is important okay so the algorithm for the Assembly of furniture at Ikea you may argue and many people do rather violently argue that they don't work they've missed something out they've forgotten to tell you something understanding the equivalent in a computer system is what algorithm design and algorithm analysis is about so to give you another example of that an algorithm to get from here to London so Nottingham to London I would say um noam's in the Midlands so you need to head south that's one style it's not exactly a great set of instructions so this again is the next thing that happens you have to figure out what the appropriate levels of instruction is to do that okay so that it's more of an instruction rather than a a vague suggestion right um so let's try again so going to London okay um bus to the train station get on train get the train to S pankas get off so that's closer to algorithm but you could also say get a taxi get in your car drive to London get out you could say fly private helicopter picks you up outside this building flies you directly to S London and drops you off now all of those algorithms all achieve the same end they get you from here to London but they're all different aren't they all a bit and some of them sound a bit Preposterous as I've said them as well but they all have two properties that I think are important to think think about okay how much do they cost cuz obviously renting a helicopter going to be pricier even today still pricier than getting a ticket on the train um driving with fuel prices um will have another cost and and so that cost is important and they'll also have different speeds one would envisage the helicopter getting there faster um than the train and the train probably being a bit bit slower than the car so each of those kind of costs and performances are important okay and so a lot of what we do is analyzing cost and performance how much resource do we have to use to do the algorithm and then how fast will it be or what will it achieve or how will it do that a lots of what we do in Computing is to actually figure out the best algorithm the best set of instructions and the best tactic for those sets of instructions to do that so things like sorting a set of numbers becomes important not that you eventually get a sorted set of numbers but that they're done fast enough and that the cost of doing it is low enough