News from PANUG/BizNix - October 30, 2003
http://panug.org - http://biznix.org


LINUX BOOT CAMP
Weve moved the venue for the Linux Boot Camp so there's
now space for more people. Details are here:

http://alcpress.com/training/

Click on Current Schedule of Classes.


MAPPING IP ADDRESSES TO COUNTRIES
by Steve Coan

The following Web site provides a service to locate the
country an IP is assigned to.

http://ip-to-country.directi.com/


The PERILS OF CLOSED SOURCE SOFTWARE
Submitted by Rich Sheppard. Here's an article about
Appgen going belly-up, stranding users and customers.

http://panug.org/53


MIGRATING TO OPEN SOURCE
by Robert L. Brown
Network Manager 
U.S. District Court

The IDA Open Source Migration Guidelines provide practical
and detailed recommendations on how to migrate to Open Source
Software (OSS)-based office applications, calendaring, e-mail
and other standard applications. These guidelines have been 
designed to help public administrators decide whether a
migration to OSS should be undertaken and describe, in broad 
technical terms, how such a migration could be carried out.
They are based on practical experience of a limited number of 
publicly available case studies, and cover a wide range of
management and technical concerns.
See:
http://panug.org/54

Why Open Source Software/Free Software (OSS/FS)? Look at the
Numbers! This paper provides quantitative data that, in many
cases, using open source software / free software is a reasonable 
or even superior approach to using their proprietary competition
according to various measures. This paper's goal is to show 
that you should consider using OSS/FS when acquiring software.
This paper examines market share, reliability, performance, 
scalability, security, and total cost of ownership. It also has
sections on non-quantitative issues, unnecessary fears, OSS/FS 
on the desktop, usage reports, other sites providing related
information, and ends with some conclusions. An appendix gives
more background information about OSS/FS. You can view this
paper at http://www.dwheeler.com/oss_fs_why.html


HEAD START RAFFLE
by Keith Rosen

I work for Early Head Start Family Center of Portland, a social
service agency that provides free and low-cost day care for
low-income families. I am running an on-line raffle in which
entrants can win an eMac computer. Donate just $3 for one chance
or $7 for 3 chances to win a brand new eMac. All donations are
tax deductible and benefit Early Head Start Family Center of Portland.

For more information or to Enter Raffle:
http://www.ehspdx.org/Get_Involved/Sub-Pages/Donate_Raffle.html 

Visa and Mastercard accepted. Questions or to verify validity:
raffle@ehspdx.org or call 503-236-9389.


FEEDBACK
Jeremy Grand responds to the October 21 Coordinated Spam Attacks
article by Ed Sawicki:

SpamBayes (and other client-based spam detectors) use the
technique of scoring each email based on past experience. That
is, they learn from the user what is spam and what is not. This
is highly accurate and effective in my brief experience, and I
think the principle could be applied to scoring sources for
inclusion in black lists and white lists both. If an ISP would
collect identified spam from individual users it could build a
database useful for training a spam filter. The whole thing
could be automatic. I believe the technique works best when
there is a database of good email too, but that is something of
a privacy issue.

Ed responds:
The techniques that you describe - often called Bayesian
filtering - certainly do appear to work well - for now.
However, they just offer temporary relief from spam. Spammers
are already compensating by mangling the text of messages so
the filter can't pick out text. One such technique embeds
legal HTML tags and HTML character entities in the text. This
mangling increases the size of the message many times.

Sure, the anti-spam folks have developed techniques to
decode the HTML before applying the filters and this will
work - for a time. The spammers will compensate by further
mangling the messages using techniques that we can only
guess at now. Suppose, for example, spammers send their
messages as images - further increasing the overall size of
spam messages. Do we then try to retrieve the text using OCR?
If we do, our mail servers will need quite a bit more
horsepower. Maybe we'll be back to mainframes just so we
can fight spam.

Note also that Bayesian filtering must receive entire messages
in order to "learn" whether messages are spam or not. This means
that the spammers have consumed your bandwidth. Some of us are
concerned about stolen bandwidth - not just spam messages
appearing in inboxes.

I think techniques that rely on examining message content will
never be a reliable way of blocking spam. The answer is in
better protocols for moving mail and authenticating senders. 
Until we have these solutions, it's easier to identify
white mail than to identify spam.


DISCLAIMER
PANUG and BizNix welcome contributions from all members.
Member contributions do not necessarily represent the
official positions of PANUG or BizNix. The views of
members that contribute frequently may appear to be the
official position of the group(s). If you contribute,
you'll be adding vital diversity of opinion and outlook
to these broadcasts.