Re: New Theme proposal & other stuff

["Aaron D. Turner" <aturner@best.com>]

-----BEGIN PGP SIGNED MESSAGE-----


I'm back, causing trouble as usual...  Just ask Mike! :-)

On Fri, 18 Dec 1998, Dana M. Diederich wrote:

> > On thing I should add is we need to decide on a non-popular subject that
> > will impact the design.  Ad banners or no ad banners?  I know no one
> > *likes* banners, but $$$ is a good thing when you want more hardware.  
> > VAR and other corporate supporters won't always be so ready to give
> > something for virutally nothing. Having $$$ in the bank *before* we need
> > it is a good thing.  I for one don't have extra cash laying around that I
> > can dontate to the site.  Setting up a not for profit org isn't that
> > difficult.  Also, a lot of Linux/OSS related sites now have ads: lwn.net,
> > Linux Today, Linux World, slashdot.org etc.
> I'll speak up about this issue: I'm certainly with Roger regarding money and
> advertisement.  If this system is properly designed, it'll be distributed.
> Roger's paradigm is accurate: hardware is nothing.  Content is everything.

I would say that is too extreeme of an opinion to take, but reasonably
valid.  As you and I discussed on one IRC session, a distributed solution
is exponentially more difficult to do correctly than a monolithic site.
It is not an impossible task, nor one that we can not overcome; but is
there an easier way?  As someone said recently, "If you can get me a PII
400Mhz, I can write the SQL queries."  Money, and the hardware it can buy
can solve difficult problems.

> Speaking for myself, I can, within a week, get access to thousands of MIPS
> and hundreds of gigabytes worth of machine just by asking some people.

Now you tell me!  Where were you when I was on my knees begging? :-)

> If this project gets visibility, you'll have more offers for space and
> computer power than you'll know what to do with.  If the project doesn't
> take off, the resources aren't needed in any case.

No doubt.  I still get people offering to mirror the RHLUFAQ even though
I've taken down the request for mirrors.  I turn them down, not because I
don't need more mirrors (who can have enough right?), but because the more
mirrors I have the more headaches I get trying to manage them.  After so
many, it's no longer is worth it.

> When linux.org was down not too long ago, I offered via e-mail to host them
> on my site.  The reply, much delayed, was that they had been totally
> overwhelmed with similar offers.  Many many OSS minded people have resources
> to give away on the 'net.

You may want to email them again in that case- they've made a request to
get some mirrors up before 2.2.0 is released.  The point I was trying to
make is that we have the abililty to get really fast boxes with really
fast connectivity.  While there's a lot of people who are willing to
donate server/bandwidth, few people have the resources to do so for such
demanding sites such as kernel.org.

> This isn't about weather it's 'right' to run advertisements.  You're correct
> in saying that OSS certainly allows, even encourages, that sort of thing
> in many situations.  But unless a project NEEDS money, the project should
> not worry about trying to get it.  The reason is this: money complicates
> the free software process, a lot.  Who's account does it go in?  What
> bank?  What do we spend it on?  What if we don't agree on what to spend it on?
> What if we make a LOT of money?  What if we split into splinter groups,
> who gets the money?  In the same way money can turn a friendly card
> game into a very unfriendly experience, money can also pollute the best
> intentions and good wills in a free development project.

That is a *very* valid statement, and one that I don't take lightly (I
hope nobody else reading this does either).  Obviously (to me, right this
second, and subject to change without notice) we'd have to do things like
create a non-profit org, get an account with some bank (who cares which)
in its name.  IMHO, the core developers would have control over the $,
each with 1 vote or something like that.  Their job is to be responsible
leaders and we need to trust them to do so.  If any of us feels otherwise,
then we have a serious problem that extends far beyond that of money.

Regardless, I don't think that a Linux website that generates an income
has as many potential problems as a group of people developing software.
The difference is that writing OSS software is expensive in man hours but
cheap on capital.  People generally already have a computer and all the
software they need to write their "product".  A web site however has
certain expenses beyond the developer time: server hardware and network
bandwidth being the most expensive.  Anyone who thinks we are getting a
server and bandwidth for "free" is mistaken- we are already "selling
advertising" in the form of a logo with the phrase "This site is
powered/hosted by ....".  I can tell you without even asking NaviSite that
they're "giving" us around $15-20K worth of service in a 12 month period.

> Right now, at no cost or obligation, this project has a good chunk of
> Roger's machines available, as well as my commercial server.  At the drop
> of a hat, both Roger and I can get even more resources.  I suspect other
> members of this list have resources at hand as well.  Why not concern
> ourselves with the aim of the project, rather than the big hardware?

I agree, we will have little problem getting machines located at other
sites to help mirror content.  I see a few problems though:

1) What and how do they mirror?  We've been talking about using Ht://dig
(which now has boolean searching btw), and MySQL.  This doesn't translate
into a mirror that Joe can setup in an afternoon.  Mirroring the static
HTML but not the search engine(s) is kinda pointless.  This means that we
need not only quality hardware, but quality people who can fix things when
they go wrong.

2) How do users access the mirrors?  Reality is that 90% of the users will
go to www.linuxkb.org and ignore the mirrors.  And when the server goes
down, they won't be able to look up the mirror list to access one of the
backups.  Round Robin DNS to avoid this "bookmarking" problem is a
sub-optimal solution in many ways.

3) There are administrative headaches with having mirrors not 100% under
our control.  They stop being mirrors for whatever reason, don't update
regularly, etc.

All these issues are resolved by having multiple servers under our control
at the same physical location.  I'm not saying that these and other likely
problems are unsolveable in a distributed architecture, but the solutions
generally require significant tradeoffs.

My personal goal from the start of this is to create a "site that not only
looks professional, but performs professional".  Hence, my working to get
quality hardware/connectivity.  Having mirrors in the traditional OSS
sense does not fit that goal IMHO.  (I'm interested in hearing if others
feel the same way.)

I think Dana is right when he asks: Do we want to worry about money or the
site development?  It's just that I think some of the "site development"
problems (such as distribution) can be solved easier with money and that
over all the potential problems money brings are outweighed by the
benifits.

Lastly, thank you Dana for a well thought-out response which was free from
flames and the like.  I hope everyone else who has an opinion will respond
(pro or con).  It's important that we decide on this matter in a way that
we can all live with, because like Dana says, it can be very devisive. 
Speak now or forever hold your peace! :-)

Regards, 
Aaron

- -- 
Aaron Turner           | Either which way, one half dozen or another. 
aturner@pobox.com      | Check out the Red Hat Linux User's FAQ Online!
www.pobox.com/~aturner | http://www.pobox.com/~aturner/RedHat-FAQ/
All emails from this account are PGP signed.  Lack of a signature is "bad".
PGP Key fingerprint = FB E1 CE ED 57 E4 AB 80  59 6E 60 BF 45 1B 20 E8



-----BEGIN PGP SIGNATURE-----
Version: 2.6.2

iQCVAwUBNntX0zM3jpXy1kJtAQGqmgP/dG0to3cfZMH+cIaLJu3CS5rrrnF8dbJl
Mb/WJ0n1KXcYEDKnaklaDxMrStUxILQQX43nc3OM9qgT6MVrP3PwZOIXRK9EBLvy
Z4g0a0ruFK2bbKSpsjRrzghiQTJ+pkjiNfm2GFnnpO9S2ZfluFM1bDfBa3sxHku7
o8XGkIWU+S8=
=/RIf
-----END PGP SIGNATURE-----