RE: ZLG (Automation in phlog discovery)
=======================================
Ze Libertine Gamer has challenged me with some interesting thoughts on the
recent "crisis" about SDF's phlog listing. This phost is a reply to ZLG and also
a broader explanation for some folks who I think have it all wrong.
Let's start with this. ZLG writes
"[psztrnk] seems to think that there isn't a reliable way to syndicate phlog
updates. I'd argue that there is a semi-reliable way to do it."
Yes, except no. You do realize that you are not arguing, but literally proving
my point, right? "Semi-reliable" is not by any means "reliable". If you order a
fruit basket and you got a basket full of veggies, that can be considered a
"semi-fruit basket" because, well, the basket, but the contents are simply not
what you asked for.
Let me break this down. Since you mentioned that you are not an SDF user, let me
explain what the "phlog" command did.
1. First, it announced that you use your gopherhole for blogging, and
2. second, it got your gopherhole listed in a separate gophermap accessible from
gopher://sdf.org reserved specially for gopherholes which feature a phlog.
The concept you described in your phost is bad by design and it fails right at
the beginning. Here's why.
Technically it can work,
1. if we assume that all gopherholes are or have phlogs (not true) or
2. if we just want to list gopherholes that have modified contents (not what we
aim for).
For such a design to work, you must hand-pick gopherholes that in deed have a
phlog and then look for modifications. Congratulation, you just re-created
Bongusta.
Even you yourself start with talking about "automating phlog discovery". And
that is the point. What you then describe is just a change monitor for *all*
gopherholes setting aside the question whether *they are phlogs or not*. How can
you possibly automate the process of picking only the gopherholes with phlogs?
You look for a folder called "phlog"? I have a folder called "log".
You look for a folder called "*log"? Many just dump entries to their roots, or
use totally different folder names like "notes" or "scrapbook".
There is no naming convention for either posts or folders and you just
simply can't create a model that covers all use cases or that can set aside user
behavior.
And this is why the old listing and Bongusta could (can) work: it relied (rely)
on human intervention. Bongusta is curated. Phlogs are manually added to the
list, therefore it in deed has phlogs only. Former SDF phlog listing required
users to "announce" that they have a phlog on their gopherholes, so they can be
listed in SDF's phlog listing.
Both uses manual opt-in approaches, although in one case it is triggered by the
users themselves (by entering the "phlog" command), the other requires Logout to
add the phlogs by hand.
And this is the filter you can't automate. There is no algorithm that can
reliably decide whether a gopherhole has a phlog or not. And when you write
about "automating phlog discovery" then you talk about "automating gopherhole
changes discovery", which, you are right, *can be automated* to be
semi-reliable, but, and let me emphasize this one more time, this is not we aim
for.
We want to list phlogs, and phlogs only, like before.
Since you mentioned RSS and Atom: these examples are totally unrelated to the
issue. In a feed collection (eg. your OPML) feeds are hand picked by the
user. And the user sees what he or she subscribes to. It is basically Moku Pona.
And since we already have Moku Pona, that is not an issue any more. The issue is (and
remains) the lack of a central entry point to the SDF phlogosphere. Which was
previously present, but it is gone now.
And since you asked me to give some details on my idea, he it goes: I'd like to
have what we had before. Either a combination of self-announcement and a changes
monitor, or just a self-announcement system where you have to scream "phlog" two
times to get you listed.
But without either hand-picking or self-announcing actual phlogs, you scream for
Betelgeuse and all you got is a bloody unicorn.
