วีดีโอ

I would love to add "what you should refute with" advice to the program. If we've already shown a card, show it again. If we're pretty sure they've already deduced a card, show them that one. Otherwise, show them the card that reveals the least information (I haven't thought at all about how to measure that) I too have a hunch that it's best not to reveal a room unless you absolutely have to. Make people walk around the board as much as possible 😈

If you know, you know

So smart! And now you know, in case you didn't already, which data structure you're using if you wanted to write this as code

😂 and don't forget the weight complexity. That last stack was heavy!l

Find the place in the video where I too use "stack" to mean "generic tower of plates" when I'm referring to a generic tower of plates that's actually a queue The dream is two dishwashers, so you never have to put any dishes away again. But then you're forced to use every single dish before you can run it, so.... 🤔

Hahaha hilarious 😂 and thank you!! I happened to catch that clip, and a funny video idea just fell in my lap. I couldn't *not* make this Awaiting other commenters to see if I taught well enough...

@@LetsGetIntoItMedia You did!

A very thoughtful question, thanks for asking! I decided to take out the explanation to keep it tight. In fact, I thought the same at first. But the numbers do come out different doing it in a single pass, and it has to do with doube-counting different permutations of scores Ex. should "FG, TD, FG" count as the same thing or different from "TD, FG, FG"? This video implicitly treats those as the same thing. And it turned out, that was everyone's intuitive decision as well. I didn't clarify that in any interview, and we all just assumed order shouldn't matter. Either way is perfectly valid, as long as you're clear whether how you want to count permutations. Doing it in a single pass counts those as separate things, so the numbers come out higher. Ex. for a target score of 5, the answer is 1 if you don't care about order (field goal + safety) or 2 if you do (FG+safety and safety+FG). In fact, you have to be careful about that in the tree too. You can't branch to scores smaller than the one you came from. That is, once you follow the 3pt branch, you stop branching to 2pts. It's worth trying it out to really see it. The single/multiple passes through the paper strip is the same thing

Fun little secret: the morning of the interviews, I had to scramble to update a bunch of things when the paper system numbers were coming out different from my code, and it was exactly this distinction 😆 I was doing it in a single pass by hand, and my code wasn't. The fun fire drills of launch day 🏈

@@LetsGetIntoItMedia Ah ok that makes perfect sense. I hadn't thought of that ordered-vs-unordered distinction. Thanks so much for the detailed answer!

✌️

Yeah, the interviewees were so nice and so fun! I enjoyed bantering with them, and our little friendly competitions 🤜🤛

Haha thanks! And it worked in the moment to get us to a conversion!

Thank you! It was a fun thought experiment in the moment, and even more fun to see how nicely it worked out in practice!

Thank you! Dimensional analysis is so fun and useful 🤩

Haha! We are so untrained in metric, that there are only so many touch points for intuition. Ex. for kilometers, I literally have to think about running a 5k or 10k, which I know are 3.1 and 6.2 miles. That's the only way I can get an intuitive sense of a kilometer. So when traveling, if someone says, "oh it's only 2km away" I have to think "that's under half a 5k, so less than 1.5 miles, so yeah! You're right it's not far" 😆

@@LetsGetIntoItMedia Yeah, I agree. Metric is great for lazy scientists that don't care how far away the object actually is, they just need the number. Imperial is the human scale. It's obvious what size is in feet and inches and miles.

@@ExistenceUniversity I don't know if I agree about *that*, just that Americans have more exposure to imperial units so we're trained to be familiar with them. But between feet and meters, pounds and kilos, liters and ounces/quarts, they are all on a roughly everyday/human scale. Except for Fahrenheit, which uses a wider range for everyday weather and therefore has smaller gradations, which I think is more useful in day-to-day life (unless you use a decimal place for Celsius, but that's annoying)

Yeah I was pretty amazed how close it was. Also, lucky to pick wine instead of 2L soda! Then it would have been harder to estimate how much soda was left over

I am Canadian, so this is just background noise math, 1 gal is about 4 litres. So if 1 wine bottle was 1 L, then we'd only need 4 bottles. But since each bottle is missing 1/4 of a L. Then you need 4 (1/4) to make up the difference which is 1 extra bottle. Ergo ~5 bottles of wine per 1 gal of milk. (It's obvious more than that because the last bottle is also 3/4L, but in terms of whole units, 5 is the max.)

Good question! I *believe* that the garbage collector will clean up anything (on some schedule) that does not have a pointer to it in scope. So a purely inline "return [...arrayToCopy, lastItem]" will stick around as long as you have a variable pointing to it. This now has me wondering about closures - perhaps its possible to have a reference to a closure that has an internal variable that's never used again, but the garbage collector doesn't know that. I'm not 100% sure about these things, so I hope someone else can chime in with a more complete answer!

Thanks! I think that makes a lot of sense. Now that I have some distance from these videos (I finished all of these like 2 months ago), I have that sense too. The concepts and "story" should be more prominent than saying and showing every detail exhaustively. Great feedback!

​@@LetsGetIntoItMedia Consider the composition of every frame of the video, Make people look at the important things. Keep things minimal. Not too many things on screen. A great example video would be: Zig in 100 seconds by fireship, or any other recent programming video by fireship. Thank you for these videos!

Thanks for the feedback! That's helpful to consider. Perhaps there's a way to highlight the code better, or show things more simply... 🤔 I'll think on it, and see what feels right for this format

Oh wow, thank you for the kind words! I am not a data scientist, but it does feel like functional programming is a great fit for a lot of those data-crunching pipelines... because they are indeed pipelines! Data comes in, gets transformed many times over, and data comes out. Seems like a perfect fit. Re: FP vs. OOP, the word I don't think using a class immediately makes something OOP. Sometimes the lines between these paradigms are fuzzy, so it more depends on how you *use* your classes. To me, it's more about "who" gets control of behavior. In OOP, classes define the shape of data (which fields exist on a class), the actual values of those fields, and the functions that are allowed to mutate that state. In FP, they can be more separately defined: the shape of data might be an interface, and the functions define how interface A gets transformed into interface B. Then whoever *calls* the function is in charge of the actual values being modified. OOP is about hiding the mutation of values inside a class that only allows transforming it in certain ways, and FP is about defining the transformations (functions) separately from the values. Hopefully that kind of makes sense? EDIT: *Actually, the next video has a small code snippet illustrating this difference* I don't know Python very well, so I won't be of much help there. But maybe the above paragraph can give some intuition on the "mindset" difference between FP and OO.

Aw thanks! Glad to hear it. I'm excited to get out the rest over some time! Hope to see you around 😁

Thanks! Glad you enjoyed :-)

Thanks, AI!

That's a good summary! I could have done this video in 1 minute 🙇‍♂️

Oooh I'm not familiar with those at all, so I'd have a lot to learn about those first. We'll see! There's so much to go before we get to that section of the series - things could certainly change. I'll keep these in mind for sure! And if you already know those libraries well, please share some of your knowledge in the comments! I would be super happy to learn from you too 😀

@@LetsGetIntoItMedia unfortunately I am aware of the existence of these libraries (they have been repeated several times in various materials on functional programming). The problem with these libraries and all others for functional programming is the same: they do not/cannot teach functional programming. They are created for people familiar with the paradigm, but completely abstract for novices. I hope your series will make me too join the ranks of people who can use these libraries effectively. PS - are you able to add a link to github with code examples?

Awww, thank you that's so nice! Two days ago I would have written that this comment made my year 2023. But now I can say it made my year 2024! 😁 I hope I'm able to deliver to your satisfaction! Any concepts in particular that you're most interested in? Any suggestions on the format/pacing? (If not now, you can always comment on a later video with questions and suggestions 🙏)

Well said! I hope that message does come across. It's definitely not necessary to know the math, or even enjoy it, to be good at FP. But as a math nerd, I do enjoy the math, and I've gotten some additional value out of it. Namely, learning to spot patterns like "this is a Semigroup, so I can simplify by combining multiple things into one", which has improved my code design. Hopefully, I can inspire others in this way, to have that additional tool for the mental toolbox. Or perhaps satisfy/spur someone's curiosity about the math. Or just entertain 🤞😀

@@LetsGetIntoItMedia I agree!

Thanks! Me too 😁 Let me know how the format and pacing works for you. I'm happy to take any feedback to improve future videos. Hope to see you back for the rest! ...just keep in mind that it'll take a little while to get to the meat of the concepts. The next episode is just an overview of syntax, and THEN we get into the basics (immutability and purity), then the declarative array functions (map, filter, reduce, flatMap), and then finally the very FP concepts like currying and all the types. Partly, I needed to spread them out to buy time to make the next episodes, partly to tweak the format/process, and partly to lay down the foundational concepts first 😁

Cool! Yeah, it really came up naturally in the order that I was able to grasp the concepts while learning FP. It even helped me retroactively to break it up this way. I'm really excited for the rest of the episodes, because I've found a code example that will live throughout the series and can eventually end up as a client-server-database application with concurrency if we want, but will start very simple. Of course, there will be some smaller isolated examples along the way too Thanks for watching! Hope you're excited for the rest :)

Glad you liked it!

Haha amazing, than you! I hope I'll be able to convince them that FP can at least sometimes be a better fit for some problems. At the very least, having multiple ways to think about coding problems is just more tools in the mental toolbox

I'm excited for where the series is going too!

Glad you think so!

Dual wielding psychology and logic is a winning strategy for so many things in life. It's a powerful combo! Use it responsibly 😎

​@@LetsGetIntoItMedia Yeah, I kinda just ended up describing how playing a game works lol. I just didn't want to give away my winning strategy but you already know it

That just shows you are employing your tactics very well 🙇‍♂️

Absolutely! Nice suggestions. I'd love to come back to these feature ideas, particularly "what does everyone else know" since that could translate directly to "what card is best for you to refute with", as you say. Regarding probabilities, it seems intractable to get a deterministic probability if you include all the suggestions (there are way too many possible arrangements of the cards, and even for refuted suggestions there are too many possibilities for what the refuter might have). I'd love to hear if you have any pointers!

Glad you liked it! Monocles are the key to success in any endeavor. Don't let anyone tell you otherwise 🧐😉

You're absolutely right! This is a great example of how whiteboarding an algorithm can hide inefficiencies like this. It's actually one reason I like whiteboarding before writing code, so that you get the overall thing *mostly correct* first. Then, when you go to write the actual code, you can spend more time thinking about optimizations like this one you highlighted, and less time trying to make the thing work. I'll keep this example in mind if I ever make a more philosophical video about the benefits and (as your comment highlights) the pitfalls of whiteboarding. In this case, I wanted to keep things simple for the video, but we should absolutely write it your way if we were actually shipping this code!

Interesting! In that case, I think it's a fine balance. On one hand, all the greatest people in their field did so by having laser focus. But for most normal people (i.e. me), there is some value in learning multiple skills, because it A) can help you cross-learn, and B) give you multiple hobbies/activities from which you derive fulfillment, which is a good strategy to mitigate risk in case one of them is taken away for some reason. I.e. it's a balance between being a jack of all trades and master of none, versus not putting all your eggs in one basket

@@LetsGetIntoItMedia Yeah, I definitely think magnitude isn't exactly the right function in that case. I think that's because it fails to take into account that spending time increasing a vector component has diminishing returns.

Yeah. Another metric to throw in is the area of the square formed between the origin and your location. That incentivizes maximizing well-roundedness. But now I'm just taking the fun out of life by over-analyzing 😂