I'm guessing that I do springs incorrectly. By the time I know I have to retreat I'm already moving up the ladder and can only make it back to safety if it's the absolute "worst" spring for the one I left on.
So... after "take off" do you guys have the entry spot (not first bounce) memorized to allow for retreat, or should I not be starting up the ladder until the follower lands at a safe spot for when I took off? Would I have enough time to climb if I did that?
I feel like an idiot dying as often as I do on the most survivable level in the game. I expect to die on barrels when I group a little too long and a fireball runs up on the right super fast. I expect to get screwed on pie factories once in a while. And I expect to get screwed on rivets. But not springs!
These are excellent questions. You are not necessarily doing anything "incorrectly", it's just a difficult portion of the game to get through until you've had a lot of experience and practice.
It is generally possible to retreat once you are already on the ladder, but the timing is pretty tight and if you are slightly wrong about judging the spring (should have kept going up instead of retreating) then you could get hit during this process.
The most precise way to judge the spring locations is based on where they bounce upon the yellow segment of the girder. However, if you do it that way it will require excellent reflexes to be able to make your decision and react quickly enough to avoid being hit by the spring (but it's certainly possible and is how most of us start out doing it). That's because you are not necessarily using all of the information available.
To answer your next question, yes, I take advantage of the information gained from the entry arc in addition to the bounce location and I believe most of the best players also do this as well. Basically, if I am approaching the ladder, I will notice (mostly with peripheral vision) if the trailing spring is "generally bad" (maybe spring numbers 12 - 15 or so) because these tend to have a noticeably high entry arc. By using this information, I am generally able to determine that I do not want to complete the ascent before I ever step onto the ladder (although occasionally I still have to retreat from the ladder). Once I know this, I will watch to see where exactly the spring bounces along the yellow section in order to more precisely determine which spring it is, which helps with the next portion of my method.
Lastly, I use a "positional based" instead of a "timing based" approach. This basically means that I take advantage of the fact that the "safe spot" is basically a sliding window of safe locations on which to stand depending upon which spring is bouncing overhead. So, if the trailing spring is determined to be a number 15 spring (narrowing it down first based on arc angle and then by using the subsequent bounce location), I know that I won't be climbing the ladder. But, I don't have to retreat all the way to the yellow block either. I only need to retreat a very short distance -- stopping and standing pretty much tangent to the ladder -- because I know that the path of a 15 spring will miss me there. The safe spot window has temporarily stretched to the right. (See the diagrams above). Once I've determined where I can safely stand a new spring will be entering and I am now evaluating this new spring's entry arc. If this entry arc feels safe (0 - 11 or so), I will use the previous 15 spring to start my approach from my new modified safe spot and finish the screen.
I find that using this positional based approach has several advantages. I find that I can create a more accurate approach than I can by trying to "time" my approach from farther away while the spring is various distances away from you. When you are essentially standing one pixel to the left of where the spring could hit you -- you simply wait until the instant it passes over your head to start your approach and you will get a nearly perfect "jump" every time. Also, you can approach from the very same spring that you were just retreating from which often results in a favorable combo of springs without losing very much Bonus. But lastly, I find that the timing of the decision process described above tends to separate out the required actions a bit so that you are not trying to do two things at exactly the same time (trying to corner onto the ladder and judging the trailing spring's bounce location tends to happen at the same time and it's hard to look in two places at once).
There is no easy substitute for experience and practice, but there are two obvious suggestions to make to accelerate the process. First, thoroughly read and absorb the information in this tread. Next, set up a MAME savestate and practice the screen in a low pressure environment. If you did nothing but practice this savestate for an hour every day within a week or two you would see big improvements.