squeak laptop for the rest of us (was: ASqueak challenge)

Jecel Assumpcao Jr jecel at merlintec.com
Thu Apr 14 21:30:27 UTC 2005


Steve Sanderson wrote on Thu, 14 Apr 2005 10:39:52 -0500
> FWIW: I saw Jecel's comments on chat yesterday, where he suggested some
> organized effort to get squeak on the $100 laptop.
>
> Count me in.

Great! We have to define what "in" means...
 
> Shooting from the hip: I suggest the 1st thing to do is to understand how we
> can help, i.e.
> 	- understand their vision for how the system is to be used;

Nicholas Negroponte has been writing about this since at least 1993
(though some of his articles about digital TV are actually more relevant
than others with more obvious titles):

http://web.media.mit.edu/~nicholas/Wired/

Seymour Papert mentioned the subject in his 1992 book, "The Children's
Machine: Rethinking School in the Age of the Computer". The context was
how much it would cost to give each student a computer instead of making
them share. He probably also looked into this in his earlier
"Mindstorms:Children Computers and Powerful Ideas", but it has been too
long since I read that one for me to be sure.

> 	- understand what's in progress currently

At a talk in May of last year, Walter Bender showed some slides about
"new ideas at the Media Lab" and the last one was a laptop with a
projection screen. No price range was mentioned. It was obvious they
hadn't yet thought too much about it because the drawing showed the
projector where laptops normally have a trackpad, so putting your
fingers on the keyboard would create huge shadows on the screen :-)

The article that started this thread did have a nice picture of a rear
projection screen. And they have secured three funding sources at $2M
each. My impression was that as recently ago as the Davos meeting (where
Negroponte first pitched this idea to world leaders) there was nothing
concrete, yet. So I would say there has been progress. And now they have
an actual time table.

For an older proposal just like this, see Alan Kay's "A Brief History of
Smalltalk" (http://www.squeakland.org/Smallhistory.pdf) and check out
the picture on page 13 called "Pendery Paper Display Transducer" Design.

> 	- then see how Squeak may fit in.

That is what I suggest we figure out and tell them.
 
> I don't have the contacts there, can someone suggest where to start?

The http://laptop.media.mit.edu/ page says:

    PRESS INQUIRIES:
    Alexandra Kahn
    Media Lab Press Liaison
    617.253.0365
    email via our contact us page

    OTHER INQUIRIES:
    Nia Lewis
    niav at media.mit.edu

but I suppose you are talking about who here would personally know those
involved in the project. Someone (not a Squeaker) has already told me he
tried to email Negroponte about this without a reply so far.

>Jecel wrote on Monday, April 04, 2005 11:22 AM
> I am very interested in this. It is much nicer to collaborate than to
> compete, even if that means having to make changes. Anyone else?

I was far too subtle above, so here will be annoyingly detailed. But for
those who just want to know what this is about without reading further:

======
How about forming a team (in the spirit of the recent community organization) to put together a proposal to be presented to the MIT Media Lab $100 Laptop group explaining how Squeak could improve their project? I will be doing this anyway, but would rather represent a group than just present my own interests and opinions.
======

A little history: I have been working on this sort of thing myself, starting with a "Logo computer" in 1983 and with "Smalltalk machines" from 1984 on. This is a long time to devote to something without a result like a selling product, but sadly in this community that situation is not unique to me. My focus has been exclusively on Brazil, though others are interested in taking my work to India and other countries. Obviously there is a huge overlap between my goals and those of the Media Lab. My philosophy is "if you can't join them, beat them!" so rather than try to get people to pay attention to me instead of them, I would prefer to find a way for us to work together. Only if that is not possible for some reason *then* I would have to find a way not to let them get in my way.

The same thing applies to the Squeak community. It is easier for me to take the lessons learned and start over with a simple "Neo Smalltalk" throwing away the historical baggage that sometimes gets in the way of progress. Like I enjoy saying, "you can't *add* simplicity". But in this case compromise is important - I guess you have to know when to set aside the great Hurd to get the ugly Linux rolling.

Here are some possible alternatives:

1) they are building a Linux machine and talk about 1GB of memory, so certainly Squeak can run "as is" in practically any configuration we might want.

2) Linux can be used as a hidden layer on a machine where Squeak is the only thing the user deals with, like in Dan's SqueakPC http://minnow.cc.gatech.edu/squeak/3502

3) Squeak can run on the raw hardware, as has been done a few times before (Mitsubishi M32D, Interval, SqueakNOS, .

4) A special processor can be made for Squeak, similar to the JOP design for Java that I mentioned in another thread.

Obviously each option is very different in terms of the time and effort required, as well as the benefits that would feed back into the Squeak community. So would this be a strictly volunteer effort like the other teams? Would there be some funding from the Media Lab or some other source? That is an important issue.

Option 1 hardly requires any effort at all. And probably is easy to have people accept it. Since users would have access to things like Firefox and OpenOffice the gaps in Squeak's application suite would hardly be an issue. Performance probably wouldn't either. The best model for this probably is the Spanish Extremadura project (and so SmallLand).

The main danger is that Squeak would only be a small (and easy to ignore) part with a lot of duplication. So you have to understand how Linux deals with N, and also how Squeak deals with N. It will be tempting to just learn the first, though it is hidden away in compiled binaries and can be studied with limited tools at best. Squeak will just be making a large system even larger.

Of course, there is no reason for the alternatives to be mutually exclusive. I know the goal is to sell minimum lots of one million units, probably all identical, but at least in the initial trial runs it might be interesting to try multiple options.

There is practically no difference between 2 and 3, though the former requires less initial effort. I have been trying to use only Squeak applications as much as possible and feel there would have to be a major clean up effort for them to be practical as the only option on a machine. Here is where a volunteer effort might fall short.

One very interesting thing to have for these alternatives would be Bryce Kampjes' Exupery compiler for Squeak. The numbers he has been posting here are very impressive (I don't know what processor his machine has. It would be interesting to see the effect on something closer to the proposed laptop). This would encourage more development in Smalltalk and less Slang/C tricks.

About a year ago I looked into 4, but would have to check some things to see how practical it could be. Even a FPGA implementation have be interesting - $2.50 for the whole computer except memory for the volumes they are talking about vs $10 just for the processor.

Perhaps this bottom up approach is the wrong way to go about this. We could discuss a "Squeak distribution for the rest of us" which would work equally well on any of the four alternatives and leave such low level details for last.

I am waiting for other opinions,
-- Jecel



More information about the Squeak-dev mailing list