A Plausible Promise of Abundant Educational Resources

Fri Mar 9 09:11:34 EST 2012

Dear Python-List:

If you find Wikipedia useful and can see the value of collaborating on a
project which could make learning material as freely and spectacularly
available as Wikipedia does reference material[1], please read on.

Background:

In November 2009, I began learning Python with the objective of trying to
understand the following research question in my PhD research study of open
source software development: "We have built Wikipedia and other big,
successful open source projects. How can we do it again (and again)?" The
critical issue for me was how to start a project which would grow to
self-sustainability. So for over two years now, my research method has been
to try to actually start such a project. I failed. Over and over. The data
collection period for my study ended.

Before I could start writing up my PhD, however, I saw an answer provided
in February 2012 by the founder of Craigslist, on Quora. When asked, "How
did Craigslist gain its initial traction?"[2], Craig Newmark, Customer
Service Rep & Founder wrote:

- from beginning, did something simple and useful
- from beginning, began cycle:
  -- asked for community feedback
  -- did something about it
  -- repeat, forever
-- got lucky with simple site design

So I have now tried to take what started as a horribly over ambitious
desktop application[3], combined with an equally inept Android mobile
application (done in Java)[4] and boiled down the core text-to-speech
functionality into something "simple and useful" which runs on the web. The
project is now called SlideSpeech[5] and it does a couple of pretty
interesting things. Interesting enough to attract venture capital.

The Future:

As of 23 February 2012, SlideSpeech Limited is a company. But it is still
an open source software development research project with a Pythonic heart.
I can now pay for development of some professional quality software by
people who know much more Java and Python than I do. Perhaps this will help
the project reach self-sustainability, although, as Derek Sivers points out
in a memorable TED talk[6], just initiating or leading a project like this
is not what makes it a success: there must be joiners and followers for a
project with a plausible promise to grow to realise its potential. The
followers are the real heroes.

Now Peter Diamandis just gave a TED talk entitled, "Abundance is our
future"[7] which everyone should watch. He talks of 3 billion people coming
on-line this decade. I want SlideSpeech to be useful and helpful to those
people. Not to make money from them but to help foster a global
conversation and global distribution of knowledge. We should all share in
Educational Abundance. Diamandis says, "we're going to hit 70 percent
penetration of cellphones in the developing world by the end of 2013."
SlideSpeech can plausibly promise to deliver free, interactive learning
material to smart phones anywhere on the planet this year. The current
working prototype does this today.

In its simplest form, the system works like this:
1) you start in your presentation software, adding a voice over script in
the speaker notes of each slide
2) you upload your presentation to SlideSpeech
3) on the SlideSpeech server (which can be your own PC, running Python,
made viewable to the world using PageKite[8]), the presentation is
"decomposed" into slide images and text scripts. The scripts are fed into a
text-to-speech engine. The resulting audio files are wrapped in HTML with
their corresponding slide image files. Finally, a link to access the HTML
is e-mailed back to you
4) you open the link on your mobile phone's web browser and the
presentation you were writing just moments before "delivers itself" on your
phone ... or on any smart phone, tablet or web browser, anywhere. No need
to organise a venue, send invitations or get people to actually physically
show up to see and hear your talk. You can just forward the e-mail.

Cooler still,
5) if you have a native application play the script using the phone's
text-to-speech engine, you don't have to download audio (or video), so you
save 75% of the bandwidth. Multiplied by billions of people, multiplied by
the number of downloads[9], that is a huge savings.

The compression of content into the simple combination of images and
text-to-speech scripts allows SlideSpeech to realise part of the One Laptop
Per Child vision using "talking" smart phones and tablets. Students can
learn on their own. Current prices on the cheapest Android 2.2 gear which
can deliver SlideSpeech content here in Auckland, New Zealand are under
NZ$150.

A Revolution in Learning:

Think of SlideSpeech as a platform like Khan Academy[10] with the exercises
integrated into the presentation. The scripts can be created and improved
collaboratively, like Wikipedia articles, or cloned and customised for
particular audiences. Unlike Khan Academy videos, the text of SlideSpeech
presentations is search-able and machine-translatable.

With presentation feedback built into the system[11], a newly created
presentation can be distributed and then dynamically critiqued and revised,
so later viewers see the corrected and improved version. It is remarkable
how quickly changes can be made, especially in contrast to the feedback
cycle a typical instructor goes through to improve their teaching.

These ideas and technologies are not new. Text-to-speech, in particular,
has been around for a long time. Lately, however, the voices have become
"Avatar-quality"[12].

These ideas are disruptive. The current working prototypes of SlideSpeech
may appear to work poorly, underperforming relative to established
products, but as Clayton Christensen explains[13], having the ability to
meet an underlying need in a radically different way transforms industries.
I like to make an analogy with trains and cars. Current educational systems
are like trains: everyone has to go between the same fixed destinations at
the same fixed times, like it or not. SlideSpeech-based learning is like
having a car: you get to go learn whatever you want, whenever you want,
wherever you are, at your own pace (which is sometimes very, very fast!).

Content is King:

Having lots of content available will accelerate the adoption of
SlideSpeech above all else. Over the coming weeks, as the system matures,
you can start preparing by authoring presentations with speaker notes. Once
the system is fully available on-line, or downloadable to your PC, you can
drop your presentations in and get out "self-delivering" talks in HTML or
even video format, voiced by a wide selection of computer voices in English
and in many other languages.

Please feel free to jump in with suggestions, contributions or forks of the
code repositories listed below.

Let's make Educational Abundance happen with Python this year.

Exponentially yours,

John Graves
PhD Student
AUT University
Auckland, New Zealand

Founder and CEO
SlideSpeech

[1] The English language version of Wikipedia has 50 times as many words as
*Encyclopædia Britannica<http://en.wikipedia.org/wiki/Encyclop%C3%A6dia_Britannica>
* http://en.wikipedia.org/wiki/Wikipedia:Size_comparisons

[2] http://www.quora.com/How-did-Craigslist-gain-its-initial-traction

[3] http://code.google.com/p/open-allure-ds/

[4] http://code.google.com/p/wiki-to-speech/

[5] http://code.google.com/p/slidespeech/

[6] http://www.ted.com/talks/derek_sivers_how_to_start_a_movement.html

[7] http://www.ted.com/talks/peter_diamandis_abundance_is_our_future.html

[8] http://pagekite.net

[9] average for Wikipedia is about one article per person per day (18.1
billion pages / 482 million unique visitors in January 2012)
http://stats.wikimedia.org/reportcard/

[10] http://khanacademy.org

[11] http://code.google.com/p/slidespeech/wiki/WorkflowDiagrams

[12] I particularly like Loquendo's Veena, Indian English voice
http://www.loquendo.com/en/demo-center/tts-demo/english/

[13] http://en.wikipedia.org/wiki/Disruptive_technology
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/python-list/attachments/20120310/7f989d7d/attachment.html>