I feel like the knight is worth more than the bishop, because of forks and stuff, and knights are very tricky pieces! Which makes me think of this question.
As you gain experience, I think you'll see that 3 is closer to being correct. However, if you want to use 4, nobody's going to try to stop you. 
A better way of thinking of it is that bishops and knights are equally unequal. They arguably have different values in opening, midgame, and endgame, how open the board is, where the pieces are located at any one moment, if they're in pairs or alone, depending on the attack or defense being used, or the individual's strategy, and so on. It's generally agreed upon that all of that averages out to a very close to equal value, but again, perhaps for some the knight may be effectively closer to a 4. As Escher said, it'll probably change as your skill and strategy evolve.
I grew up being taught that the knight was 3.5 pts and I live by that. If you think about it, they have the ability to change the colored squares and when placed well cause much more of a threat.
Your light squared knight is usually worth half a point more than your dark squared as a general rule of thumb.
And the light squared rook is worth a full point more than the dark squared one. As a General rule of rum.
In bullet, rule of thumb is that knight is between 4 and 6 points. Jerry AKA Chesnetwork sacrifices rooks for knights fairly often in bullet for no compensation just because tricky knights are much better than boring predictable rooks. But with enough time to calculate? I don't see how knights are more valuable... most grandmasters prefer bishops because they are more mobile. On an empty board they can move from any square to any other square in just two moves.
So you think knights should be worth 4 points because you always get forked by them?
I find that rooks are actually really useless in a lot of middlegames. The main reason to avoid losing the exchange is because if you lose the exchange, your opponent can simplify down to an endgame where he has a rook for a piece and THEN he is winning. They can be decent for attacking things in the middlegame, but usally they have to be clumsily lifted and moved around to files, and it's very easy to kick them out of important areas. They have pretty much no ability to attack and defend at the same time do to the nature of the way they move. But they are so unbelievably good at pushing passed pawns that we give them a value of 5 points and almost always avoid trading them for bishops or knights. I think it's for a similar reason that you can say "bishops are more valuable than knights" even though it's so situational and how, like rooks, people generally don't trade bishops for knights unless they have compensation.
Keep in mind that the two bishops have value too. While a bishop might have the exact same value as a knight, two bishops is more valuable than two knights and two bishops is more valuable than a bishop and a knight, unless the position is really closed or specialized in some other way.
Since my king has not been equiped with catapolts, knights remain the closest thing to an air attack that he has :) and they defend and threaten both colors on the board.
Knight and bishops are worth the same, depending on the position one can be better than the other.
Intelligent horses must be sick of how "knights" always get all the credit for their chess exploits...
You're right Zig. I am sick. smiles.. The whole game depends on "depending on position". That's the name of the game, aka chess! It's up to the player to develope the strength of the piece. That being said, before ever moving a piece, I see more potential in a knight than I do a bishop. Hence, my 3.5 pts rating.
Should knights be 4 points instead of 3?
No.
I feel like the knight is worth more than the bishop, because of forks and stuff.
I like vanilla ice cream more than chocolate.
Kauffman's value is that the bishop pair is worth a half pawn bonus, on average, for players rated at least 2300 (IIRC, may have been a little higher). A single bishop is, on average, the same as a single knight.
Your light squared knight is usually worth half a point more than your dark squared as a general rule of thumb.
Hopefully you ignored this post.
I grew up being taught that the knight was 3.5 pts and I live by that. If you think about it, they have the ability to change the colored squares and when placed well cause much more of a threat.
The way they teach beginners is to observe the max number of squares each piece is able to control in the corner, vs on the side, vs in the middle. Bishops are more mobile and control more squares. Knights can eventually influence any square.
Kauffman's calculation is that knights lose a little value on average, and bishops gain a little value on average, as pawns are removed from the board.
---
Lots of people said depends on the position. That's correct. Kauffamn calculated on average the difference between a knight and a bishop is 1/50th of a pawn, a difference in value that he notes is statistically meaningless... they're worth the same.
https://www.chess.com/article/view/the-evaluation-of-material-imbalances-by-im-larry-kaufman
I heard this today while watching a non-chess video, the speaker used the following to illustrate a point and it seems to fit in this convo; "Deep blue cannot calculate the knights fluctuation in value as the game progresses, a human can"
I feel like the knight is worth more than the bishop, because of forks and stuff, and knights are very tricky pieces! Which makes me think of this question.