(I'd like to do something similar)

How precise do you need to be?  Have you considered Theramin-style antennas?  http://en.wikipedia.org/wiki/Theremin

Made my first LED cube, using code from http://www.stationinthemetro.com/index.php/archives/2008/01/19/makedc-led-cube-workshop/ and building instructions from http://blog.makezine.com/archive/2007/09/make_a_pocket_led_cube_we.html

I did a 3x3x3 but I believe the Ardunio would support 4x (pins 0-13 plus all 6 analog pins).  Will put up a picture when I get that going!

Why start with lesson 0 and not with lesson 1?  :

If you open a box of chocolates, the first chocolate you grab out of the box, it's chocolate 1, not chocolate 0...  

For the rest, keep up the good work  

(aside from the zero-based array argument...) Step zero is usually the step you have to do before you can do anything real.  In the arduino instructions, it's getting set up.  In the chocolate box, step zero is "open the box".

To "start" using the arduino, all you need is the board ($35 or so) and a USB cable.  With that you can make the LED on the board blink (yay!).  Then add your own LEDs, a button or two, turn them on and off.

To use sensors, chips, or interface with most other components, it's really a lot easier to put them (and any needed other components) on a breadboard.  I like the proto shield offered by www.adafruit.com; you do have to solder it together but it's pretty simple.
http://www.adafruit.com/index.php?main_page=product_info&cPath=17_21&products_id=51
Add tiny breadboard:
http://www.adafruit.com/index.php?main_page=product_info&products_id=65
A set of jumper wires (or three) is really handy; so the starter kit isn't that bad a deal:
http://www.adafruit.com/index.php?main_page=product_info&products_id=68

I'm interested in the opposite question.  What do I need to get started with max/msp?  Do I have to buy the environment?  I guess I can try it for 30 days for free, but I'm afraid I'd like it and then be obligated to spend the $500 (ouch!).

Any special reason you're not just using perfboard?  It works fine for me with "production runs" of 1 to 10 items.

I'm about to start working on a project that involves a long string of LED's used to light up the keys on a full piano. I plan to control what keys are lit up using software on my PC. This will be my first Arduino project, and I've looked at tutorials like this:

http://www.arduino.cc/en/Tutorial/ShiftOut

For driving many LED's. In this care I'll probably have around 50-60 LED's I want to control. Am I best daisy-chaining these LED shift register chips or is there a better way? If I should use these chips, where's the best place to buy them? (Are there any free samples)?

Thanks so much.

Rather than making a chain of 50-60 of them (why not 88?) you'd be better off creating a matrix.  The easiest way is probably to consider each octave a column, and each key within an octave a row.  To light a key, make row high and column low.  This is similar to the "LED Cube" strategy.  That way you'll only need (rows+columns) pins/bits - 12+8 for a standard 88 (7 octaves plus).  That's 20, so you'll need a chain of three 595s, and three bytes/shifts.

Shift registers are cheap - like $1.50 for 8 bits.  You might also consider a MAX7219,  though it'll only drive 64 (8x8) bits.

You're going in the right direction, tho.

I have an idea for the first project we can start testing how it works collaboratively: i2c micro-shields and new i2c libraries to make its use "end user" friendly.

...

Like?  Dislike?

Like.  What do you mean by "micro-shields"?  I've used them for Blinkm and the Nunchuck (there's a guy that makes Nunchuck "shields")  I'd like to see them have long pins or long female pins.

Not sure if this is relevant, but I'd like it if standard "shields" passed through all their pins, used or not (or maybe null the used pins?), so you could stack shields.

I like the idea of having a set of forums here - maybe a category of forums - for "group projects".  Would this be a useful way to get started?  Forums: proposed projects, projects in process, completed projects?  Or proposed, hardware, software, etc?

I'm to the point in my own project (Array of Optoisolators - http://www.arduino.cc/cgi-bin/yabb2/YaBB.pl?num=1204135553/6) where I'd love to collaborate with others and get some prototype boards built.  More specific forums for this would be great!

These projects are beyond the scope of an Arduino tinkerer, so they shouldn't be mixed in with, say Freqently-Asked Questions!

It sounded to me as if he was talking about an actual piano (where the only wires are the ones that go "boing" when they're hit with a hammer!).  I saw one of these on a streetside garbage pile the other day, but didn't have a car I could pick it up with.

In which case my strategy would be to devise some sort of momentary switch that's triggered by striking the keys - if the hammers are intact, you could use those as well (and it'd look better and be classier).  Apparently velocity-sensitive keyboards use two switches, and time the difference to determine strike speed/intensity - but that's another level of complexity.  The switches themselves are the critical component, because they'll affect how the piano feels to play.

Get yourself a synth keyboard that works (yard sale, swap meet - no more than $20 for a starter; usually a few of the keys or the key buttons are broken and you won't need those), tear it down and trace where the keys go.  If you can duplicate/wire into this you don't even need the Arduino since the keyboard chip does all the matrix scanning as well as synthesis.

Wire the piano keys to match the synth matrix.  The (low-end Casio) synth's I've looked at use an 8xN matrix so key 1 would connect pin 1 to pin 9, key 2 would be 2 to 9; key 9 would connect 1 to 10.  8x5 is 40 keys and in the keyboards I'm using this is typical.  (The "voice select", volume, etc keys use similar matrix connections).

Failing that, build yourself a trebuchet.  You'll need several telephone poles, and a Volkswagen for a counterweight...

Could you use a simple spring switch?  I've seen them in a bunch of devices I've torn apart.  Basically it's a bare solid wire or long machine screw, inside a spring that's attached at one end.  When the two touch, circuit is connected.  Vary or bend the spring or attach weights at the top to control sensitivity.

See also:
http://en.wikipedia.org/wiki/TM_31-210_Improvised_Munition_Handbook

(From the "when all you have is a hammer, everything looks like a nail" department...)

Optoisolators?  I'm using PS-2501-4 but there are lots of others.  The input side works like an LED, the output side works like a simple switch; there's no electrical connection between the two sides.  I've found that the output side may be directional/diode-like, though, so watch for that.

I wanted to like Making Things Talk, but I still don't think I do.  There's a lot of useful information there, but it's buried in projects that require sensors and equipment that I don't have; and the projects themselves are really sort of stupid.  I suggest finding a copy and reading it first before investing.

I'm still in-between looking things up online and wanting to own a book.  I still buy a lot of reference books (for electronics as well as web programming, my "real job") but I end up relying on "the Google" and forums like this on for most of my info, since they're both a lot bigger and more up to date.