120 comments

[ 537 ms ] story [ 2699 ms ] thread
that's really cool, but the font contrast is killer (bad) on certain parts in firefox nightly. I signed up; I can never seem to get as consistent with chess play as I'd like. I'd be thrilled if this helped me achieve those goals.

example : http://i.imgur.com/cqNPcgf.png . The light blue is really difficult to focus on using a laptop panel. I found myself cocking my head to see it more clearly. It may be more the font than the color. I don't know. The more I look at it the more it's the grey that bothers me rather than the blue. I can't put my finger on it, but something hurts readability for me.

Yep, that's the one thing about the site that needs improvement.
Thanks for the feedback! We're looking into ways to improve the typography so that it's clear universally for our users.
The site is not helpful, and gets in the way of teaching. I am the target demographic, and went straight to the exercises. So much clicks on every step, the buttons to click on are grey and mouse focus turns them even more dull. Provide some visual clues on what buttons to click next, going forward and going backwards button both are the same color. Exercises themselves are more trivia then help in teaching what the piece does or where you went wrong. And "moving past the basics" exercises are http://i.imgur.com/c54R08y.png I just left on the first exercise itself. One of the hints: Attack two pieces at once. Yeah, right. Nothing like codeacademy at all.
The how-to-draw-an-owl example is not accurate at all.

If you would not have gone "to the exercises ignoring all the other stuff", you would have gotten a clear 6 minute lecture where Andrew Ng explains, in detail, the position in the first exercise.

In the video, he introduces three tactics, the fork, the pin and the skewer, which are methods to attack two pieces at once.

If your comment was just plain sarcasm, then I apologize, I didn't get it.

[please read edit!]

I think GP comment is right to some extent. I went in with the "moving past the basics" and watched the whole video, it didn't teach me anything I didn't know.

I found the setting up of the positions distracting - it should either have been set-up in advance or he should have talked about the positions, named the openings and such, or kept quiet. You're watching moves but they don't relate to what he's saying until the position is set and we start the actual lesson.

The grey "move on" button was not at all obvious, standard "go" green would help.

When you complete the exercises it simply prevents you moving the last piece rather than indicating on the board that the exercise is complete and you did it. It doesn't look like there is more than one possible response either, which means that additional hints could tell you the next move they're expecting. The examples were facile until the "combining tactics" - then I did Bf4 (bishop from c1 to f4) and the exercise just sticks ... did I do it or is it glitch, there's no feedback. Should I click to move on? There's no affordance or visual indication as to what to do next. Similarly there's no explanation - AFAICT - why you did right [?] moving to f4 (protects c7 for the knight to do the king-rook fork?).

The quiz questions get the feedback correct, they show the answer in green (don't know what they do if you get it wrong ;0)) but the last question then goes to a further question "undefined", what now? Should I click the greyed out arrow to move on?? Is part one over?

The board: when learning usually you get piece move highlighting wherein you click a piece and it tells you the available moves via highlights on the squares. That's usually paired with ability to click the destination rather than dragging the pieces which can be a bit awkward.

Oh, just went back to first exercise to see if it did "move highlighting". The pawn when dragged over e3 changes the "undo" to "move on", now I can't see if there are any other valid moves. Perhaps I twitched, did I get it right? Which square was the right one?

Logged in, slight problem as I had js disabled, but easy to handle and not unexpected. Takes me on to "Tactic 14954" - I've done a move (Bf5) and board is frozen, did I win? Same as with the exercises. Is the board waiting for the server to tell it the next move, is there a js engine that's computing the next move?? I'm assuming it's done but the only available button says "give up", so ...?

Aside: Is there a way to export chess problems easily to a local chess board? That would be great, could use the videos, export to a local board with extra features and that would let you play on the positions and such.

It's a good start. Looks generally nice but there are some pretty big UX problems that cross in to genuine bugs in utility IMO.

[Using FF31 on Kubuntu]

Edit: Apologies, it appears to be a problem with FF31.0 for me, Chrome is working, giving feedback and such. Will let comment stand.

Thanks so much for the feedback! This comment is immensely valuable in helping us test cross-browser compatibility. Glad to hear everything's working smoothly on Chrome.
In the videos before each set of exercises, I try to lay out all of the key ideas so you have in your arsenal everything you need to successfully complete the lesson. I probably don't succeed in this goal all of the time, so your feedback is very useful for me in rethinking how a user progresses through the exercises. Providing an option to read the solution after a certain number of incorrect moves may improve the overall experience and discourage brute-forcing the answer.
One thing I've learned in writing coding exercises, is that engineers tend to jump straight to the example code, ignoring all previous instruction. It might help to build your examples with the possibility that your student totally ignored everything you've said up until they started moving pieces.
Good idea - we might prompt the user to go back and review the video if they seem to be struggling with the exercises.
I think the main problem is that when you get into the exercises that require combinations of tactics, it assumes the student has the capacity to make that leap, or recall on demand everything they have learned and put them all together without a helping hand.

I'm towards the end of "Diving Deep" and I'm ready to give up as the exercises continually lead me to brute force moves until I hit the specific move you are looking for.

This is fantastic and a resource I wish I had growing up. Very beautiful design too.
I remember seeing this somewhere at Startup School NY. Very very impressive. I'm not a regular chess player myself but I have visited Chesscadamy a few times out of interest since then.
Great stuff. Beautiful feeling to the site, though the 'Train' section could do with an explanation as to what you're meant to do on the exercises ('tactics'). I'm tempted to suggest integrating this with en.lichess.org, a similarly awesome chess-resource.
An onboarding process for "Train" is in the works - right now for the beginner player, this section is rather overwhelming and he or she probably would have no idea what to do. The objective is to find the best move in any given position but that information isn't very obvious.
Awesome idea and implementation!

But: it seems to pre-populate the username from fb login on signup, but forbid characters that allowed in it (i.e. "."), causing a messed up situation in which you are half signed up.

Thanks for the report! We'll look into this issue.
I seem to be stuck on the first tactic (not in the chess sense). I complete it, hit the green "Next Tactic" button, and it just gives me the same one again! It's Tactic 20628, which is the first one I was given after telling the site I'm an intermediate player. Reloading the Train page puts me back on the same tactic too. Any ideas?

Loving the site so far apart from the above problem, though! Getting back into chess was one of the items on my to-do list for 2014. The Learn resources are looking like they'll be pretty useful. Great job!

Report this tactic (using the exclamation mark button) and it should give you a new tactic. Thanks for the feedback!
On exercises where one really struggles (yes, I'm super stupid when it comes to Chess -- I can play Go which is supposed to be harder but for some reason anything beyond the utter basics of Chess eludes me) it's very frustrating to be shown the same hint over and over again when it's irrelevant or unhelpful (e.g. "Think about the name of the tactic - the windmill... it goes around and around!" Yes, I'm thinking about it! Now give me a hint!)
I think the problem here isn't so much that the hint is repeated as that it's needlessly vague. You may have already solved the problem but in case you didn't I'd like to give you what I think would be a good hint because windmills in chess are actually really cool.

A windmill in chess is a tactical scenario where a the king is repeatedly revealed to checks by the same piece. A "revealed" check occurs when a piece moves out of the way allowing another piece behind it to attack the opposing king. This scenario allow you to move the piece freely while your opponent can only move his king. Here's a great example of a game with a windmill (and actually one of the great games of all time) The first check of the windmill occurs on move 19.

http://www.chessgames.com/perl/chessgame?gid=1008361

I think repetition is definitely part of the problem. A vague hint is fine the first time, but it should get more specific if you keep struggling.

Also, some of the hints only apply to the first move (or first two). If you find that move but get stuck later on, it's irritating to be shown an irrelevant hint. In this case I got the windmill started but for some reason the last move was in a blind spot for me -- though when I went back an hour or two later it was completely obvious (but that doesn't make it less frustrating when I was stuck).

I found that hint very helpful actually.

"It goes around and around"

Huh, I should repeat something.

It was helpful the first time. But obviously once I'd tried "going around and around" and hadn't found the move in ten attempts, it obviously wasn't helping any more.
Go is harder for computers. It's computationally more complex, but it doesn't mean that it's supposed to be harder for humans (our brains work differently, some tasks - such as face recognition - are trivial for us, but for a machine are way more difficult than playing chess with Kasparov's strength).

How do you even measure what is harder for a human?

One way to measure it could be the length of a chain that looks like this:

"Person A can beat person B with 90% probability, but Person C can beat Person A with 90% probability."

You'd want to limit the people you count to being, say, tournament players with stable ranks, such that e.g. their rank was not altered by the games against the last ten people they played.

My intuition is that, even if we stick to US players, Go ends up with a chain that's a few multiples of the Chess chain in length. This intuition mostly comes from Go's handicap system, where there are many levels of "I can give 9 handicap stones to a worse player and still win, but a better player can give 9 handicap stones to me and still win."

But I could be totally wrong. I've never heard of anyone try to measure this distance before and it would be awesome to get some pointers to such a study if it already exists.

Perhaps the absolute value difference in Elo rank between large communities of Go players vs. Chess players does the same thing, depending on which Elo parameters you use?

(And, uh, to get back on topic -- nice site! I'm doing some problems now.)

The best NBA team will beat the worst NBA team with extremely high probability. Much higher than the probability that the best MLB team beats the worst MLB team. For example, when the San Antonio Spurs played the New Orleans Pelicans on March 25, the average odds implied a probability of around 96%[1]; when the Texas Rangers played the Minnesota Twins on August 31 2013, the implied win probability was around 75%[2]. I think that your system would probably score basketball much higher than baseball (if I had to make a guess, there are probably 5-6 90% spans between high school basketball and NBA basketball, and 3-4 spans between high school baseball and MLB baseball), implying that basketball is a harder sport.

It would certainly be a very interesting metric, and one I would love to see calculated more formally. That said, I'm not sure it captures the intuition of "hardest game" that effectively. To be the best at a game depends primarily on the number of people who play it at the elite level and the time investment they put in, combined with the importance of natural talent. Suppose there were a game which had even longer chains than go, such as if each player always defeated the next stronger player with 90% probability, and there were 1000 players who played recreationally. We wouldn't think of this as a particularly hard game, because to be an elite player at the game, all you'd have to do is take it on with a professional level of commitment and you'd quickly move up in the ranks. So the difficulty of a game at elite levels must at some point include a hefty weighting of popularity and commercial appeal. As to the importance of natural talent in this idea, you simply can't play in the NBA if you are 5'0", and having such a physical limitation would tend to increase what people perceive of as the difficulty of the game. If there were a board game that required a 150 IQ to even understand the rules, people would definitely call it a very difficult game.

1: http://www.oddsportal.com/basketball/usa/nba/san-antonio-spu... using the highest odds and converting to probability, which is equivalent to the "best bid-offer spread"

2: http://www.oddsportal.com/baseball/usa/mlb-2013/texas-ranger...

Very good points

Chess is the best researched game of all time, yet it still progressed tremendously in recent decades (thanks to becoming a professional sport in the West - around the Fischer era - and the appearance of computers, of course). We know how long the "food chain" is, but we can't tell to a certainty how long it can become (how close can a human get to perfect play), even if the pool of motivated players is very wide.

Current world champions are standing on the shoulders of the giants. Alekhine or Capablanca, geniuses as they were, wouldn't have a shot for the title if they didn't catch up with modern developments. And there are PLENTY. Memorization plays a big role these days. In one of the games in the Topalov vs Anand match, Anand slipped because he messed up the order of moves - all home preparation!! - around the 27th move or so. (Does this mean that humanity has already traversed the entire steep part of the learning curve?)

So the chain is still growing. Does it mean that chess is becoming more difficult game? It's a Platonic question, really :) Do we mean chess as such, and the innate difficulty of the game, imprinted in its rules that remain fairly stable - no major rule changes for the last couple of centuries, I believe? The IDEA of chess, so to speak?

Or do we mean chess AS IT IS PLAYED, or the reflection of the perfect idea :)

In the early era of competitive chess we had the so-called Romantic period, in which aesthetics of the game, the style, was valued over cold efficiency and - for example - it was considered ungentlemanly not to accept a gambit.

Was it just a cultural convention hurting the level of the game? Or was it a different flavor of the game, in which a player had to fullfill an additional responsibility they are now relieved from, and thus not quite the same discipline?

Just my two cents, some chaotic food for thought :) Sorry about the language, not a native speaker here.

Unfortunately, there's no option to "give up" just yet on exercises. We're looking to implement a feature where after an incorrect number of answers, you're given the option to read the solution. It's probably less frustrating this way than forcing users to attempt to brute-force the exercise.

Qualitative personalized feedback for content is in the works - optimally, we'd be able to tell you why the move you gave was incorrect and why the solution is best. This is an interesting technical challenge because it requires us to cover a large number of possible solutions.

I think I'm a pretty good chess player but a lot of the times I'm just playing the game without tactics or strategy but I win most of the time. Do tactics actually help against a great player
Well, it's easy to know how good you are in chess. What's your rating? If you mostly play friends&family, and they don't play much chess, just having played more will be enough to win.
If you are a "pretty good chess player" you are most certainly playing with tactics and strategy, you are just not aware of it. Becoming aware of it would make you a better chess player.
I have a friend who got me into the whole chess scene he plays online against computers and online chess players all the time he is not a professional or anything but pretty clever chess player I learned from how to play the game and now I can beat him pretty easily took some time though. But I notice people do things like Queens gambit or other tactics. When I play I play so I don't lose a bishop for a knight basically point based strategy and just try not to get any of my pieces taken out but I've never actually studied any tactics or anything like that maybe I am playing with it just no aware of it.
Using the point system works at a basic level, but beyond that it is not so helpful. A bishop isn't always worth more than a knight, a bishop pair isn't always great, a rook isn't always worth two minor pieces, etc, etc. Knowing how to make these judgments is part of strategy.
You're probably using tactics in your games even if you don't realize it. In many cases, winning material is the result of playing a tactic. At the beginner/intermediate level, almost all games are decided by tactics and/or blunders (i.e, your opponent losing a piece). Increasing your vision over the board offers up more opportunities for you to play combinations against your opponent and avoid falling into traps.
Saw Andrew Ng and thought it was for machine learning.
I'm around 1000 on chess.com blitz (5min games), and found a lot of useful stuff here. Especially the positioning is what I feel like I'm losing on at the moment. I don't do big blunders anymore, but somehow I'm often in a worse position than my opponent.

I started a few months ago, and have progressed nicely. But lately I've been stuck. Looking up resources, they have often been too simple ("learn chess, this is how the pieces move") or too advanced ("5 variations of opening X" isn't very useful at my level of play).

You may find this book helpful: http://books.google.com/books/about/Chess_Strategy_for_the_T...

It starts with basic strategy as a refresher, then intermediate, then advanced. It's nicely organized so you can skip parts you already understand.

You may also benefit from starting to study endgames. They're simpler than openings, and can help you improve your game significantly.

We're trying to structure Chesscademy such that it provides a feeling of progression. A problem that I had while I was playing competitively was to identify material that would take me to the next level instead of covering information that was too easy/difficult for my current playing ability. Each course should provide you with the requisite knowledge to complete the next one.
nice idea but i think there should be prepared board for each demonstration instead of getting it created during the video.
This seems really nice. I've played chess on and off since I was a kid, but never really studied it. Lots of the basic tactics I saw in this course I remember using intuitively. It's nice to have a formalised notion of all those things, so you can consequently move forward to the more advanced stuff! :)
(comment deleted)
Hello HN! I'm Andrew, a national chess master and one of the co-founders of Chesscademy. We're a part of this year's upcoming Imagine K12 cohort and it's great to hear all of your feedback so far. I'll try to answer any questions you guys have and respond to your comments.
This is absolutely fantastic. Well done for creating such a great product Andrew, really well implemented.
Well done, Andrew! I particularly like your team photos. How long did it take to design the site, etc?
Our first iteration of the site took 2 months to build. The three co-founders worked on it from June - August of last year and the product was well received by HN (we reached the top post a year ago as well).

Chesscademy had a temporary hiatus when we went back to school last fall, but we've been working on it full time since January of this year. The product and user base has grown substantially since then.

This is a great project, to my mind. My grandfather told me the basics when I was around three years old, but I’ve never really approached tactics, especially in such a structured, accessible and clear way you do it in the videos.

You definitely got me hooked up on this for some time :) Thanks, Andrew, to you and your team!

Thanks so much for the kind words! Happy learning!
Thank you. My daughter and I did so the exercises together and she now loves chess =].
Lovely idea, but the FB connect is a mess. When I see that I assume I don't need to create a login. Instead, I'm prompted by Chesscademy that my username (one YOU created for me) isn't a good one because it has a period in it (you made it) and I still have to enter a password...why have FB login at all?

Hope this feedback helps.

Thanks - we've realized that there are some problems with FB connect when characters like a period are used. We're working on fixing this.
Appreciate that you took the time to personally address a reported tactic. You are doing things that don't scale the right way - tactics are 90% of this website for me at the moment, and they're starting to get pretty arbitrary and frustrating. So to manually filter through them will encourage me, and others in the same boat to stick with Chesscademy.
The training section is great, I love it, and IMHO the problems presented are harder than the ones at chess.com (which I'm paying for). The only problem that I've noticed so far is that some of the training problems end in a very strange way (e.g. opponent gives away his queen or rook for no reason whatsoever). It's not a big issue, but you should try to make it a bit more realistic.
Thanks for the feedback! I'm actually going through all of the tactics now and adding motif tags and explanations. Hopefully, this will make the solutions a bit more evident.
I'm working through "Introduction to Tactics" and I enjoyed the video, but the lessons are frustrating. I'm doing a valid "fork" or "skewer", but the examples just slap my hand and say "no, try a fork or skewer". Explanations why moves are wrong would make the examples far more useful, so I understand why the very specific combo the tutorial is looking for is the right choice.
We're working on incorporating personalized qualitative feedback into the content so you're provided with an explanation of why your move is incorrect and why the solution is optimal. Specific explanations for every move is probably infeasible, but we should be able to provide reasons for the most commonly played moves.
Excellent.

Did you guys build this as a framework by any chance or do you have plans to?

Are you referring to the structure of the content on the site as a framework? We think that this approach to education has significant applications beyond chess, but at the moment we're solely focused on making Chesscademy work for its intended purpose.
A short series of hints would be much more helpful, that's great. Also, at least one of the exercises (I'm thinking of mating with a rook and the king) puts you back to the original starting positions if you make a stalemate move. A bit frustrating to have to go through the beginning all over again.

I haven't played in at least 10 years and so far it's been a nice refresher, and a good looking site as well.

Lovely website and great resource. Can I ask what you built your website with?
Great site! My first chess learning resource at the moment is Mato [0], I suggest to take a look at how his videos are: nice and smart. Another good thing of Mato is that he has not a strong english accent, so for one who is not english motherlanguage like me is easier to understand what he says... by the way: what about adding subtitles?

[0] https://www.youtube.com/user/MatoJelic

Subtitles are definitely something we want to add! It's a bit of a time-intensive process for our small team, but it's on the road map.
I wish there was a similar site for Go(weiqi)
Here's a site that allows you to learn Go by letting you perform exercises in increasing difficulty, which I think is like codecadamy:

http://321go.org/home/?ln=uk

It's surprisingly complete, I've practiced on evenings for months and I think I haven't even finished half of their program yet.

(comment deleted)
The interface looks a lot like treehouse. Is there any connection OR just an inspired design? (or neither?)
I'm gonna call that a "heavy" inspired treehouse site. Treehouse has a unique look, and this gave me the impression that it is part of that site. Same colors (or very similar), illustrated shape badges with 3 color palette, similar layout. They even have the hide slider.

It is inspired & allowed, but leaves a bad taste.

Is this yours? If it is, take a look at the buttons, they all look "deactivated". #DADFE1 is too unsaturated. Or at least, add an hover colour on it.
Thanks for the note! We'll look into improving the coloring of the buttons so that they are clearer.
What would the potential business model be?
We're looking to partner with schools districts to add Chesscademy to the academic setting. Optimally, anybody would be able to set up their own chess club using the site and Chesscademy could work well with instructors both in and out of the classroom. For example, a teacher could assign courses to students and track their progress. The in-class lesson material could be derived from these observations.

The specifics are still up in the air, and we think being part of the Imagine K12 cohort this year will provide us with the educational expertise to create a product to sell to schools.

The rating system for problems seems like it needs to add in some kind of variance factor, or at least a provisional period. I'm altering problems' ratings by 20+ points with my default 1200 rating despite having less than 10 problems completed. My actual strength is hundreds of points above 1200, and so if I play on this site without making an account a bunch of times, I'm going to artificially lower every problem's rating...
Yes, we're currently using Glicko-2 but we're not accounting properly for the Rating Deviation that incorporates previous games played (tactics attempted). This will be fixed in a future update. Thanks for the report!
I also think it would be really nice if the position switched to an "analysis mode" after completion. Sometimes, especially if one gets a problem wrong, it's nice to examine the position more deeply. Bonus points if some open-source chess engine is integrated (I presume this can be done client-side somehow?).