Etoys, Alice and tile programming

Alan Kay Alan.Kay at squeakland.org
Sat May 10 15:43:45 UTC 2003


Good questions ....

At 9:19 PM -0700 5/9/03, Ned Konz wrote:
>On Friday 09 May 2003 08:41 pm, Alan Kay wrote:
>>  At 8:50 PM -0700 5/9/03, Jeffrey T. Read wrote:
>>  >Building anything complex (and even at 15, to me all the
>>  > interesting stuff was complex) would have tired out my mouse hand
>>  > in a hurry.
>>
>>  Yes, I think this is generally true. The tile stuff is great for
>>  just starting, and especially for younger kids, but it gets awkward
>>  when it is scaled up. However, the tile version that the Alice
>>  folks did is for older kids and does scale better than the etoys (I
>>  think). Another version of assisted programming -- popups for next
>>  possible stuff -- could really do a better job these days.
>
>What gets awkward first? Is it the vocabulary growing, or is it the
>connection between the scripts, or what?

Well, I try to use the etoy scripting for as much media programming 
as possible and for testing out what high schoolers might do (in 
anticipation of a true Omniuser scripting system someday), and this 
often leads to dead-ends in etoys.

The two things that get awkward quickest are (a) the difficulties 
involved with building expressions (it was not in the scope of the 
original design to deal with more than the simplest linear 
concatenations for expressions), and (b) the constant changing back 
and forth from one viewer to another given that only one viewer is 
visible at a time. We have created various designs for UIs to go to 
the next stages of assisted scripting but are just starting to 
implement UIs for them.

A third awkwardness happens if there has been some success with the 
first two, and that is (c) the difficulties of making changes in 
tile-based scripts. For many cases, one doesn't really want to undo 
back through the hierarchy of subexpressions. Perhaps one wants to do 
something like a simple text selection to indicate the aggregate that 
is going to be replaced, and then use the DnD tiles to make the next 
expression ....

The idea of having the system try to show you what can come next in a 
script goes all the way back to the 60s and was pretty awkward in the 
manifestations I've seen -- and there have been many attempts since 
-- still kind of awkward. But I still think this is a good idea that 
hasn't found a good UI yet. The CMU folks did a nice job of trying to 
deal with the tradeoffs of endusers making Pythonesque scripts using 
tiles and menus ...

This is worthwhile thinking about, because in our 3+ years of testing 
etoys with children (and parents and teachers) there is no doubt that 
one of the biggest selling points to them is the DnD tilebased 
scripting. It just gets rid of a lot of stuff that is normally in the 
way for beginners and gives them more initial braincells to think 
about their first projects.

>
>
>Do individual scripts get unwieldy or is it the large number of them
>that does?

Yes and yes. Scott Wallace and I have come up with what we think is a 
cleaner and nicer look for the scripts that also has the virtue of 
being able to turn the tile look off and wind up with cleaner and 
more compact scripts that can hold more. This looks like it will work 
nicely for (some of) the transitions to more sophisticated scripting. 
Josh Gargus has been experimenting with DnD tile UIs for making full 
expressions including parens. Nathanael has been experimenting with 
using his Genie recognizer for recognizing hand drawn 2D expressions, 
etc.

Cheers,

Alan


>
>--
>Ned Konz
>http://bike-nomad.com
>GPG key ID: BEEA7EFE


-- 



More information about the Squeak-dev mailing list