[conspire] Real online presence v. outsourced SaaS hosting, again

[Rick Moen]

A discussion occasioned by the fact that the control-freak editor of
_Linux Gazette_ is likely to shut the magazine down and is lobbying to
move operations to Facebook of all places.  But that's not the
interesting part:


----- Forwarded message from Rick Moen <rick at linuxmafia.com> -----

Date: Mon, 16 May 2011 16:12:14 -0700
From: Rick Moen <rick at linuxmafia.com>
To: Thomas Adam <thomas at xteddy.org>
Subject: Re: [TAG] LG - what's next?
Organization: If you lived here, you'd be $HOME already.

Quoting Thomas Adam (thomas at xteddy.org):

> I guess this sort of suggestion is trying to reach LG for "the masses"
> -- as if LG needs more advertising on that front, where social trends
> are annoyingly considered a hub for anything of interest.

This seems as good a place as any to finally articulate my (personal)
principles of online presence:


1.  I do everything I reasonably can, myself, using open source on my
own machines or people and groups I trust.  Thus, I don't do my own 
street mapping, nor operate a Usenet news server, but for many other
purposes I'm a fully autonomous island of presence on my own machines.

2.  I pay for services, by preference.  If you pay for a service, you
are a customer.  If you don't (e.g., GMail), you are product.

3.  Where I must use others' services, or find it overwhelmingly
more convenient than doing the same task myself, I favour _commodity_ 
services rather than specialised ones.  That creates more chance for 
competition and for mobility to other service providers if I find one of
them unsatisfactory.

4.  For similar reasons, I eschew 'integration', and pick and choose
technology on an a la carte basis by preference.

5.  I never, ever use a service merely because of network-effect alleged
advantages.  (I make one exception to this:  I have LiveJournal login
'rinolj'[0] strictly in order to follow my wife Deirdre's friends-locked
LJ postings, which otherwise would not be possible.)

6.  I read purported contracts.  If the terms of service purport to
authorise the other party to do something sinister, my assumption is
that they will do it.[1]

7.  I regard information as having value.  If a firm want my personal
data, they are welcome to offer me money.  Otherwise, my data are not
available.  Thus my lecture and related article on Web browser privacy:
http://www.svlug.org/prev/2011feb/svlug-lecture-2011-02-02.odp
http://www.svlug.org/prev/2011feb/svlug-lecture-2011-02-02.txt
And also:
http://linuxmafia.com/~rick/keplers-2008-10-23


> For those people where social networking is their only means
> of communication (sad; I was told by my project manager the other day
> that her son thinks of Facebook as communication, as if emailing is
> too much effort, and he's only sixteen, FFS)

Here's something a bit more sad:  My guess is that one of the factors
that drive people towards Facebook and several other social networks is
that those customers see no other way to evade spam.  Using SMTP without
being overwhelmed by spam has become a specialised skill.


> However, at the risk of repeating myself, there's
> two fundamental issues that seem to plague LG:
> 
> 1.  Lack of interest.
> 2.  "Burden" of maintenance.
[...]

Agreed, in total.

I don't actually have much experience with github, but hear good things.

However, I doubt that the Neanderthal option of just assembling an HTML 
issue from a Linux server and maintained local scripts is actually very
arduous.

If I had to do an issue tomorrow and had half an issue's worth of
articles, I'd write a couple and then do some rough markup in vi with 
references to the existing LG CSS.  It wouldn't look ideal, but it would
work and wouldn't be very difficult.


> [1] Yes, that's right FBI/CIA/MI5/MI6 -- go get the *real* bad guys.
> And leave Echelon monitoring for proper illegal communication.

It's actually less the governments that are the issue than the hundreds
of corporate 'little brothers'.


[0] A joking reference to my 'winolj' essay that I posted quite a few
years ago:  http://linuxmafia.com/faq/Essays/winolj.html
(The username stands for 'Rick Is Not On LiveJournal', i.e., I do not
put any personal substantive creations there, but instead put them onto
my own Web site and at most link to them.)

[1] So, for example, I'd never have been taken by surprise by
Amazon.com's removal of people's copies of _1984_ and _Animal Farm_ 
from Kindles -- and 'accidental' deletion of students' notes on those
books -- because I'd have read the Kindle service agreement and noted
that they arrogated that right to themselves, and assumed they would use
and abuse it.  As it happens, I instead received a Barnes and Noble
'nook' e-book reader as a gift (Android device), and immediately rooted
it and updated it to a community-maintained Android distribution from
nookdevs.com, specifically so that I would be in undisputed sole charge
of it.

----- End forwarded message -----
----- Forwarded message from Rick Moen <rick at linuxmafia.com> -----

Date: Tue, 17 May 2011 15:43:20 -0700
From: Rick Moen <rick at linuxmafia.com>
To: Thomas Adam <thomas at xteddy.org>
Subject: Re: [TAG] LG - what's next?
Organization: If you lived here, you'd be $HOME already.

Quoting Thomas Adam (thomas at xteddy.org):

> Would an example of this be a hosted colo server or something?

Let me approach that question the long way.  My 'winolj' essay
(http://linuxmafia.com/faq/Essays/winolj.html) links to a Tim O'Reilly
article that I found very, very interesting, though the lessons I draw
from it are not quite the ones he intended, as I'm not his target
audience but rather the adversary of his intended audience.

  http://tim.oreilly.com/articles/paradigmshift_0504.html
  Open Source Paradigm Shift
  by Tim O'Reilly
  June 2004

This was the article that intoduced the current and consistent O'Reilly
Media (formerly O'Reilly & Associates) party line that open source is
obsolete, that Web 2.0 hosted applications / Software as a Service is
the wave of the future, and that proprietary licensing is not any kind
of problem.  (Thus, for example, the O'Reilly people claim that Twitter
is inherently a thousand time better than identi.ca and the open source
StatusNet software on which it's based.)

Anyway, the thrust of Tim's argument is that: (1) The main effect
of open source has been to commoditise software infrastructure.  (2) 
Such commoditisation has started at the low network and
OS/services/protocols levels and has gradually moved higher up the
'software stack'.  (3) However, the existence of cheap, reliable,
commodity infrastructure at progressively higher levels up the software
stack opens opportunities for proprietary profit advantage at even
higher levels.  (4) Those opportunities for proprietary profit advantage
typically involve leveraging social 'network effects' and aggregating
other people's creativity on infrastructure you control.

At that, point, Tim lapses into a protracted hand-waving bullshit
exercise about creation of an 'Internet operating system' and how he's
just shown that licensing doesn't matter.  However, ignore that bit.
The interesting part is _before_ the author starts madly waving his arms
trying to fly towards his desired conclusion.

Tim intended his essay to be a road-map for entrepreneurs in turning the
Internet into a set of proprietary profit-making fiefdoms built on top
of open source (and, in particular, seizing the 2000s' opportunity for
circumventing other people's copyleft licensing through SaaS deployment
where there is no software distribution hence no triggering of copyleft
obligation, that arose because coders were caught napping by SaaS and
haven't yet widely adopted Affero GPL and similar).  However, I
_personally_ find it more useful as a roadmap to _avoiding_ proprietary
profit-making fiefdoms.  Major points:


(1) Thus, for starters, when you hear 'You need to participate on
AOL^WFacebook because everyone's there', that's exactly when you should
say 'No way in hell', if you value the Internet as a commons.

(2) Commodity services are ones that are fungible:  ones that are
purchasable from multiple sources in a competitive market, and
preferably where you can do them using your own open source software and
data on your own machines if you prefer.

An example is StatusNet's microblogging software.  My friend Evan
Prodromou wrote it as a Twitter-compatible federated (OStatus protocols)
service running on publicly available, commodity open source software.
The identi.ca service runs on it, but anyone can run a StatusNet node.  
If were to become interested in microblogging, I could use the identi.ca
service, which appears to be run by people with some integrity
(including Evan), _or_ I can set up my own autonomous StatusNet node.
Thus, even if I choose to use identi.ca for convenience's sake, I am not
stuck using it, as everything I do there can be without problems
exported and rehosted on a different StatusNet host.

Thus, identi.ca and other StatusNet instances are in stark contrast to
Twitter, whose users are screwed on the many occasions when the
service unexpectedly fails, and will be completely SOL if/when it
suddenly goes bankrupt and shuts down like a raft of other Web 2.0
companies people were choosing to rely on.  (Twitter have carefully
avoided implementing the OStatus protocols, as interoperability is not
in their interest.)

The question that O'Reilly pundits carefully never ask, because they
know the answer and find it inconvenient towards the cause of making
money, is 'What happens if $PROPRIETARY_HOSTED_SERVICE shuts down?'
With Twitter, all existing data are lost or at best must be reworked
from local cache/backup data of some sort.  With StatusNet instances,
there's built-in full portability.

(In fairness, microblogging data are almost always intended as
ephemera, so nobody cares very much if everything before this instant
evaporates.)

(3) A la carte is to be preferred over 'integration'.  Microsoft
Corporation and others started trying to sell 'integration' as a means
to gain proprietary advantage in the 1980s, and it's always had far more
baggage than it's worth to users, but some people never learn.

For example, my local science fiction convention has, for the last
couple of years, relied on a poorly architected and inflexible
virtual-hosing setup (Webmin-administered, no root login) that at least
is built atop real open source software.  I've helped them (as
postmaster and listadmin) make it functional for the past two years,
with frequent frustrations because nobody but the vendor has root and
can install/configure the key software packages.  However, for 2012,
they're intending to move to Google hosted applications.  Why?
Integration.  They'll be able to use proprietary Google Calendar 
and Google Apps services, and they won't control either software or
data.  I wish them luck, but they won't get my help, and I'll be
sitting around waiting to help them recover from their error.


A hosted colo server (or virtual machine) is one way to use commodity
services.  It's not the only way.  You might be happy enough with your
local ISP's SMTP / IMAP services and not need to control software /
configuration.  In that case, you might have a shell account with the
ISP's mail software handling the Internet domain you own on a virtual
basis, thus giving you a user at mydomain.com mailbox and a
http://mydomain.com/ Web site.  What would make that 'commodity' is that
it's fungible with very similar service offered by thousands of other
ISPs around the world, any of which you can move to just by signing up
for service and repointing your DNS.

A proponent of Flickr might ask what I use that's exactly like Flickr.
Wrong question.  If I had a lot of photos, I might use some sort of
gallery software, of which there's a large variety available in open
source.  (The best known is in fact Gallery, though it's overfeatured
and has had a history of serious security problems.)  None of that open
source software gives the exact feature set as Flickr, which is A Very
Good Thing since Flickr itself is rather overfeatured.  For example, I
think few open source photo gallery frameworks support public comments
-- which is just as well given the difficulty of the comment-spam
problem.

A proponent of Google Apps might ask what I use that's exactly like
Google Apps.  Again, not entirely the right question.  One must decide
what functionality's worth pursuing, and at what cost.  

Proponents talk up the advantages of multiuser editing of documents, 
but it's not clear that they're actually benefiting from the limited
amount of editing concurrency supported.  To the extent that they are, 
one would have to ask whether, e.g., checking an RTF document out of
version control and then checking changes back in might not give all the
advantage that's useful.

Although it should also be pointed out, Gobby works as a truly
collaborative network-editing environment for at least plaintext,
and is multiplatform (Linux, Mac OS X, MS-Windows):
http://ddaa.net/blog/gobby-first-look#gobby-it-works

In my long experience, people who think you can usefully have multiple
people working simultaneously on a complex, formatted-text (up to and
including DTP) document are fooling themselves.  Merge conflicts or no
merge conflicts, you end up with a horrific mess that a single
experienced person then must unravel at net cost in time, effort, and
quality.

Are the sort of more-Unixey, less-software-intensive solutions I favour
less polished and require a bit more from the user?  Hell yeah.  Are
they more robust, portable, and maintainable?  Hell yeah.  After Google
Docs and all the SaaS / Web 2.0 stuff shuts down and goes bankrupt,
it'll still work.


[LiveJournal:]

> That's where it starts to break down though, Rick.  You have no alternative
> but to create such an account.  That must be annoying.

However, please note the nuances:  By policy, I create nothing
substantive on LiveJournal.  E.g., I post no blog entries (as my profile
page, which directs people to my _real_ Web presence on linuxmafia.com,
clarifies).  I don't even post anything more than trivial comments onto
friends' journals:  If I happen to write anything substantive that
requires real effort or seems to have lasting value, then I post it onto
my _own_ Web server and link to it, in order that I be able to assure
its continued existence, in order to build my own Web presence, and to
avoid sharecropping for LiveJournal.  Friends choose to sharecrop for
LiveJournal:  I have only the minimal access necessary to view their 
friends-locked journal entries, and nothing more.

> > 6.  I read purported contracts.  If the terms of service purport to
> > authorise the other party to do something sinister, my assumption is
> > that they will do it.[1]
> 
> I see this as unavoidable though.

Please note the nuances:  Amazon.com is not the only producer of ebooks.
Yes, you cannot use a non-rooted Kindle without Amazon.com being able to
control what you have on it, _but_ Nooks, for example, have no such
problems -- even without rooting them, which turns out to be easy.


----- End forwarded message -----