Hacker News Reader

> I often read papers full with math I don't understand and I am forced to go backwards, but it only works because I hit a previous knowledge not so far back in that graph.

This is the best way to learn new things (besides spending time face to face with an expert tutor). You have a motivating problem, an idea of where you are going, and a whole graph to traverse backward until you fit the pieces together. You will remember the parts you come across because they are connected together (both forward and backward), and you will be working closer to the edge of your capabilities. It is much easier to achieve flow doing this kind of learning.

If you have too many unconnected edges in the graph, then sometimes you have to give up and pick a closer/easier subject to study, but the time spent was still not wasted, as it provides useful orientation for future attempts.

Not sure how well it works if a lecturer or textbook is guiding you though. The whole point is picking your own problems and questions, and then figuring out how to solve/answer them.

> I once knew a successful person who told me "I never start a project I can't fool myself into thinking I can complete within a weekend"

Not sure about “never”, but doing many of these little projects is often a more valuable use of time than banging against a project whose scope is so big you have no idea where to start.

This is a great comment.

As an adult, the way I’ve come to learn is exactly the same as you. It’s way more efficient, too: if you begin with the end and work backwards, you can ignore the things that aren’t very useful. Like if you’re learning how to make a graphics engine and you don’t know linear algebra, you can learn only those parts of linear algebra that are relevant to the work of creating an engine and leave all the other useless stuff behind. (And in linear algebra there seems to be a ton of pretty useless stuff.)

This is probably a bad idea for a mathematician or scientist, but for a programmer who’s in the trenches just trying to build stuff, it’s great.

I had a history book in school which basically attempted to tell history backward. It didn't work very well in my opinion. The problem is, telling history backwards makes for en extremely "whiggish" view of history where everything is understood in the frame of how it lead to the current state of the world. It makes it very hard to escape hindsight bias.

I think it might actually work. In teaching "history," you are always picking an essentially arbitrary starting point. I think you could easily start with something like nuclear negotations between the Soviet Union and the USA, then work backwards from there. Why was it these two powers that had nuclear weapons? Well, let's talk about World War 2... How did this Hitler guy come to power in Germany in the first place? Well, let me tell you about World War 1... Obviously there would need to be some linearity to each individual chapter, but moving backwards in time at the thematic level absolutely seems possible and even advantageous in certain respects. History is a very different subject from the sort of applied technical things you're describing. But in fact you might be making a good argument for doing it this way: you're describing the need for some kind of anchor in a knowledge graph. There probably isn't a better anchor than the current existing world as everyday people experience it. Why not start there?

I think it only sounds like a good idea because you already know the details of what comes before. If you had absolutely no knowledge of history, I think this method would not work well.

There are so many details that led up to WW2 that going backwards would just feel messy. All the things happening in Europe would have to be described in a way where you already know about the next war but now have to list off these different events.

I don't think that going backwards is going to make things interesting for anyone who isn't going to be interested in it forwards. The technique of teaching (a good teacher) will do more for this than teaching things backwards.

I see what you mean but that's sort of what I meant by doing it linearly at a topic level. Absolutely, you don't want to start with the end of World War 2 and then move backwards to its causes. But I think it's entirely feasible to learn about the Cold War, then study WW2, then study WW1, etc... I don't know, I don't have especially strong feelings about it, but it definitely seems at least viable.

I would imagine ww2 best described in both orders: you start with the main participants, their alliances, and their general strategies/incentives, then you iterate over each country prior to the war, and why they started to get involved in it, and finally you discuss the war itself, and its major events

I think people in this thread might be overstating the degree to which history isn't already taught this way. I just got done teaching a world history class (from the Paleolithic to 1500) at a university. Granted, we don't tend to teach classes like this in a completely reversed chronological order, like Pulp Fiction or something. But I began the course with the famous "hockey stick" graph of human population and had them think throughout the class about how we got to where we are. Was farming inevitable? Were world empires? Was the Industrial Revolution? If yes, why? And if not, how might it have gone differently?

The class was oriented entirely around the starting point of our present (both as a group of people living in a university town in California and as a global society). We worked backward from that present to key in on some of the main changes of the past several thousand years, notably the origin of written language, agriculture, urbanized societies, global trade, and science. Teaching world history without that present-day hook would be virtually impossible because there is just too much to cover and you need some kind of limiting factor. As you note, the best anchor is the current existing world, so that's what world history teachers tend to use.

Yeah you're entirely right. I was a history student in undergrad and graduate school, and obviously this sort of thing does occur naturally in some courses. But I do think it's still fairly rare that someone intentionally strategizes working backward in a rigorous way. I mean again, I don't consider this some be-all-end-all, I just consider it a viable strategy (among many alternative viable strategies). Mostly I'm responding to the opinion that this couldn't feasibly work.

Simple. You pick a current event. A, and say how it was influenced by this historical event B, and how it was motivated by an even more historical event C.

The only real advantage over starting with C, is that you have the motivating current event A to capture the attention of your students.