I noticed the README comment about NNTP's limitations. I wonder if a new NNTP-over-HTTP protocol could find enough traction among the text-using crowd, or if ActivityPub could be used to provide a similar feature set efficiently.
Actually one of the reasons for writing the app was to reduce my compulsive browsing of social media. A feature like this would go in the opposite direction :) But if you really want it, feel free to tinker with the code, it should be pretty straightforward.
I'm coding something similar for nostr events. Might add HN too keep different feeds on the same TUI location. Text is less distractive than these modern crafts of devil called CSS and god-forbid: images.
In later peak Usenet, there was also threading, and scoring.
Scoring rules would let individuals define rules for how thread subtrees and siblings are ordered, highlighted, and hidden. For example, comments by a particular user you like bump up the ordering priority for that comment, and also maybe some boost to the tree above it. And some keyword you think tends to happen in threads you don't like adds a negative score to that subtree, and maybe it hits a threshold causing it to be hidden (though the positive score from that user you like participating might save it, with you thinking that maybe they'd step into a flamewar and say something smart). And, separate from rules you defined, you could also manually hit a key to raise or lower a post or subtree (which could be used as an alternative to the earlier feature to say hide a subtree no matter whether more comments are made on it, like you might do for a flamewar).
Not really retro news, since it assumes everyone is using a large window and full-color terminal.
I was hoping to use it with one of my retro machines, and got
use_default_colors() returned ERR
It should have --monochrome, and --ascii options.
The Unix ethos used to be "run anywhere on anything," but even basic programs like top assume a window of a certain size.
Similarly, Lynx markets itself as being for retro environments, but a number of its flags that would be useful in retro environments are ignored/broken.
Surprisingly (to me), htop is very well-behaved, and works even in tiny windows, or on tiny monochrome screens.
To be precise, it only assumed 80x25 terminals with 16 colors, which I think fits into some definitions of retro, but I see your point :)
I've just added support for ascii & monochrome modes, and reduced the required screen size to 80x12, which I think is a reasonable minimum for readability.
Would you mind sharing what kind of hardware did you try it on?
You're almost there. But 80x12 is still a bit big for truly retro hardware.
Remember that 80 column screens were not the majority of computers in what we now consider the retro age. 40, 36, and even 32 columns screens were common.
Also, terminals would often have status bars, reducing the amount of vertical real estate. 12 sounds small, but with status indicators active, you're looking at 10 or even 8 usable lines.
What commonly used UNIX terminals had less than 80 columns? VT100 had 80 columns, IBM PCs had 80 columns. I remember some 8 bit computers having text modes where you could choose to display less columns in return for getting more colors, but those computers never ran UNIX and getting on usenet was surely a pipe dream.
I agree, any terminal I ever used to access a shared unix host had 80 columns. Typically a VT100 or ADM-3a or something like that.
TRS-80, TI-99, Commodore, Apple etc. home computers had 32, 40, maybe 64 columns and you could get terminal emulator programs for them but most people would use them for CompuServ or America Online, or maybe BBS logins. I think I tried using my TI-99 terminal emulator with a unix host and it wasn't very successful. The host had no idea what my terminal was and operated in a very limited "dumb tty" mode.
Eh, to do it properly, 40 cols and less would require adjusting the whole layout and I just want to keep the code simple.
I'm also not sure if the code would be fast enough to run on so old machines, I mean it's Python and not optimized at all.
But feel free to fork and tinker if you're interested, at this point it's not really far off.
Why not an NNTP gateway?
NNTP doesn't support browsing threads by title (let alone paginated) and requesting their messages on demand. Clients need to fetch metadata of all available messages in all available threads in advance. Given the volume of messages on HN, synchronizing them to the gateway is not practical. Even when attempted, some clients struggle with the sheer number of messages in a single group.
That was also my first question, why not do it "properly", and while the reasons given are understandable brick walls, it seems to me that the real solution would be a re-vamping of the over-aged NNTP protocol, or even an outright new
protocol.
I always liked USENET news for its "pull" approach, and for the same reason hated email newsletters. The threaded discussions via Emacs GNUS were a delight, and modern Web-based forum solutions are distractive and never respect my font/color settings; plus, they also diff in UX.
Interesting take. I never tried to register on lobsters and never used slashdot too. But it is ironic you mention it. I tried to register and I got that
> New user registration is now approved by Slashdot administrators. Please contact feedback@slashdot.org and let us know why you are interested in registering, and what you can add to the discussion.
But to be honest I don't know about registration practices of either of them.
That's an interesting development with Slashdot, one that I was not aware of, but even so, I feel that unlike with Lobsters, you will not be asked by Slashdot for approval from a preexisting Slashdot member. I guess this is what throwaway Google accounts are for.
I read both and this has never been in my thoughts. If you're going to make such bold claims out the wazoo, back it up with some proof? Examples would be a good start.
They're known to spread harmful lies about AI, thereby spreading AI hate, in their comments. It's worse than that because their mods and admins are complicit in it.
Anyhow, the proof of restricted registration, and the groupthink effects thereof, are for all to see.
I would have to "out" certain individuals to share the evidence, and I do not feel comfortable doing it. It goes like this though:
1. New Lobster user starts posting links to valid new non-bs AI related open-source projects.
2. Established users feel threatened that their world is being turned upside down. They start posting hateful bs comments, and escalate it to outright lies about the projects. These lies are nothing but FUD that allow them to dismiss the projects out of hand. As we know, the first stage of acceptance is denial. It is obvious that the established users are not into AI, and will never be. Because these are users are well-known, their lies gather a lot of upvotes from sympathizers who don't care about fact-checking anything. New user's account is now at permanent risk due to the downvotes. For the sake of argument, the links posted were to GitHub projects with 100+ or even 500+ stars.
3. New user calls out the lies, but only gets deeper into trouble with the mods and admins taking the side of the established user despite their obvious lies. After reporting, the new user gets banned for calling out the lies. Moreover, his corrections of the lies get deleted. Nobody cares for the fact that the established user had stopped posting links long ago, and the new user could've continued to post many more relevant links if things had gone his way.
---
Whether you believe this report or not, surely you can appreciate that groupthink is undesirable for any community, and new blood providing new ideas are a must for a community to continue to prosper. In stark contrast, Lobsters' constrained registration system which allows only referred users to register is one that maximizes groupthink.
> might just as well be "growth hacker/spammer hate".
This is utter nonsense since projects in question were in no way affiliated with the submitter. There was never any growth hacking or spam as such. No commercial service was even submitted, so there is nothing to grow.
Another crazy thing that the admin is known to do is to steal credit from a submitter, replacing the story with another story that gets submitted later.
AI aside, the central theme there is that submitters are hated, not valued.
If there is one thing I have learned, it is that hate fixes nothing in this world. It only makes thing worse. Only love fixes things, otherwise separation neutralizes them.
