Weblog Without Honor or Humanity

Few ideas for RSS aggregators. I was a late convert to RSS aggregation
but now that I know how much more news I can read now (compared to my previous
method of web browsing) I’m not going back. I use Syndirella which is good but
basic so I have a few ideas for improvements.

People who read this blog also like… We live in a networked world. Aggregators
have access to great information: lists of RSS feeds that people subscribe to.
It would be fairly easy to implement the Amazon-style recommendation system:
for each RSS feed aggregator could mine other people’s lists to find blogs that
those people also like to read.

One-panel view. Most aggregators use three panels: list of feeds, list of entries
in a feed, feed itself. My regular usage of an aggregator is: launch, get all
updates, read through them all.

Automatically calculate the update interval. Syndirella by defaults checks
the feeds for updates every 1 hour and enables you to change that. Why not remember
when the feed has been updated and automatically re-adjust the update interval?

Some feeds are better than others. I like some feeds so much that I want to
read them as soon as they show up. Other feeds I would like to read only when
I’m extremely bored. Syndirella doesn’t allow me to make this distinction. I
should be able to mark feeds as important or not and in default reading mode
the non-important ones would be cached locally (so I don’t loose the updates)
but filtered out. Then I should be able to switch to "reading everything"
mode where I also see the unimportant feeds.

Some entries are more interesting than others. Most of the stuff I read is
of the "read and forget" kind but sometimes it contains information
interesting enough that I would like to save it for the future. I could do it
manually by copying the text somewhere but I’m too lazy. I would like to have
an easy way (one mouse-click or one key-press) to mark an entry as important
(and then an easy way to locate those entries).

Pepper is back. In an interesting twist, Pepper
(a text editor) is back. The story with Pepper was that its author apparently
got angry at Linux customers being impolite to him and at the same time not
buying the program (or some such; this is just a guess based on publicly available
information) and decided to stop development, there were some talks about open-sourcing
it but it required $10 k to be paid which was too much to get in donations so
Pepper was sold to someone
else but apparently not really because now it’s back and done by the
original developer. Or to be more precise: the original developer decided to
continue Windows and Unix version (strangely enough given that it was supposedly
Unix people that made him stop developing in the first place) and someone else
will develop Mac version.

Anyway, regardless of how much I like Pepper I have to say that there’s not
much money in the “text editor” business those days. There’s too much of a good
stuff available for free.

Even more PDA market data. Handango sells application for PDAs and every
quarter they publish a very rough analysis of sale trends in Handango
YardStick (pdf). This
PC World article presents PDA sales data. The trend isn’t very good: hardware
sales decline (21% compared to last year) but still they ship more devices per
quarter than the total size of PDA market at some point. PDAs are here to stay.

Palm, Pocket PC sales numbers. Q1 2003
PDA/Smartphones sales numbers for EMEA (which is Europe, Middle-East and Africa)
has been published (I would
really like to know how they collect their data). According to this data
Symbian-based devices has 53%, Windows CE 24%, Palm 19%. The interesting number
is dominance of Symbian. Here I would like to know how much of it is for pure
phones and how much for more PDA-like devices like P800.

Treo sales numbers.

According to
this article Handspring
shipped around 180.000 units of Treo in total, 71.000 in last quarter. It’s good
to know because Treo is one of those PDA/Phone combos with GPRS (i.e. you really
have internet access everywhere) gadgets and there are plenty of interesting
applications that could be written for a platform like that. But they won’t be
written if the platform is not big enough to sustain commercial software development
for it. I wonder how does that compare to
shipments of Sidekick.

Unix-haters book available on-line.

“The UNIX-HATERS Handbook” is a classic. It’s a rant but intelligent and hilarious rant.
I’ve read dead-trees version few years back and now it’s also
available on-line.

Do you read the old papers?

Alan Kay’s talk at O’Reilly Emerging Technology Conference was widely reported.
Here’s one of the most complete notes from the talk.
Read it, it’s good. There’s one thing in the talk that got my attention. During
the talk Kay demonstrated a lot
of cool systems designed more than 30 years ago. His point was that those
systems were better than what we have today. To quote:

This stuff is better than anything in our handhelds today. We could implement it
from they papers they wrote then, but no one reads the papers that were written
in the 60s.

I think he meant that no one bothers to read those papers i.e. it’s our laziness at fault.
But I found more credible reason: those papers are not easily available. For
example I tried to find Sutherland’s 1963 MIT thesis on his Sketchpad, which
pioneered interactive graphics. It’s not freely available on-line. You can buy a
PDF version from MIT library for $29 (not $6, like Kay said in the talk). I did
buy it as an exercise but the price is ridiculously high and it’s MIT’s failure
as an educational institution that they greedily squeeze every last cent from
their copyrights.

A quick Google and CiteSeer
searches for Engelbart papers also turned up nothing.

Even if those papers were freely available there’s a problem of selection.
Thanks to web we have problems with information overload. Too much stuff to
process. We need to filter and one good strategy of filtering is trusted
referral i.e. if someone we trust says that a given paper is good, we might read
it. For example I spent $29 on Sketchpad paper because Kay (whom I trust)
mentioned it. Ironically it’s people like Kay could fix this situation. Why
don’t they put a web page with recommended papers and let the blog word-of-mouth
do the meme spreading thing? I couldn’t do it because no-one cares about
"Krzysztof Kowalczyk’s recommended papers" page but a lot of people would care
about "Alan Kay’s recommended papers".

BTW: for $50 you can also buy a book
"Seminal
Graphics" that contains Sketchpad thesis (and other break-through papers).

Bayesian filtering for non-spam detection purposes.

I was thinking about using Bayesian filtering for the "news watcher" type of an
application (i.e. an application that given a keyword(s) would alert you about
new articles on the web relating to those keywords). One problem is that
keywords aren’t very good for narrowing the search result e.g. if I just gave a
keyword "Danger" because I wanted to monitor news related to
Danger, Inc., I would probably be flooded
with lots of irrelevant articles because "danger" is a very common word. From
what I understand about Bayesian filtering, if I spent some time building a
database of interesting/un-interesting articles the Bayesian filtering would be
able to pick up interesting articles based on the content by the way of advanced
magic. Of course, as is with most of my ideas, I’m not bothering to test the
hypothesis. Anyway,

this guy did used Bayesian for a text classification not related to spam.
Interesting and in Python.

P800 reviews.

All those reviews
of SonyEricsson P800 but none that will answer the really important question: is
there any plan in the U.S. that gives GPRS connection, how well does it work in
practice and how much does it cost?

Is open-source viable?

He, Cringely doesn’t think so this week.
He already admitted that next week he’ll change his mind, but I’ll save you the
wait and deconstruct Cringely. He says:

Here is the core argument: There are a thousand Open Source projects that get
started out of need or fun, are maintained for awhile for fame, then get
abandoned because there is no reason to go on. Eventually, the programmers come
to understand that “users” are people who yell at you to fix stuff. So Open
Source is inherently flawed. It only works because otherwise unknown programmers
can get 15 minutes of fame using the Internet as low-barrier entry into
introducing their skill to the world.

There are two problems with this statement. The first part has an exact parallel in non-open-source development:
there are thousands of non-open-source projects started but, for one reason or
another, only few become interesting and revenue-generating programs.

Second
problem is a sin of omission: if you accept that need or fun are the only
reasons for starting open-source projects and those projects are always
abandoned when "introverted nobodies" realize that it’s not fun at all, then it
open-source could have been "inherently flawed". When poorly thought out
hypothetical arguments contradict the facts, I usually stay with the facts.
Today the fact is: open-source is viable and growing. As long as no-one can
provide a good argument for why it should stop growing, I’ll assume it’ll grow
in the future.

Even Cringely admits that there are successful open-source projects but
doesn’t try to dig into why those projects and not others are successful. Ok, he
does try by saying that successful open-source projects are those that have
developers paid by some company but to me it looks like mixing up the cause with
the effect. Companies pay engineers to work on Linux or Apache or Samba or Gnome
because it’s a successful project, not the other way around (the only landmark
open-source project that I know of that was sponsored by a company from the
start is Mozilla).

After that Cringely goes to describe a hijacking attack on open-source. In
his script, soon to be a major motion picture, an evil entity (which is
inevitably impersonated by Microsoft) puts their engineers on open-source
project. They contribute a lot to the project therefore becoming its de-facto
leaders (this should tip you off that Cringely lives in a dangerously
imaginative world: as we all know evil entity otherwise known as Microsoft can’t
produce any good software from which we can safely infer that their developers
would be laughed out of court). Anyway, back to the Cringely plot: after
becoming de-facto leaders developers steer the development into stray waters and
make sure that this project will never compete with evil entity’s products.

Back in the reality: what prevents such hijacking is a license and the fact
that those are real humans directing those projects. It’s not like you can say:
hey, let’s just something incredibly stupid and they’ll say: yeah, great idea.
And even if you can and they do, someone can always fork from the pre-hijacking
code.