I don't like most approaches that are based directly on ratios (even though I've suggested them myself before), because those ratios are sometimes way off (for an example of this, consider Symmetrick, who now has a jaw-slackening merit-to-post ratio of ~23:1 over 516 posts, but, before much of their stuff was self-deleted, their ratio was closer to ~0.77:1; still good, but, ~30 times less so). Also, if you read the rest of this post, I think you'll see that I'm advocating for something that wouldn't continually force any particular "time frame" on anyone (neither in terms of real time, nor in terms of "time" that only advances by one unit when you make a post).

---

Agreed. But, that's not what I'm proposing. When I say, "In effect, you'll need to earn at least 1 merit for every 10 posts you write (on average) if you wish to prevent your account from slowly drifting toward a lower rank.", what I mean is, "A system that charges you 0.1 merits for each post you make has the practical effect of establishing a rank-wise equilibrium requirement of 1 merit for every 10 posts.", but, there's nothing in what I'm proposing that can even notice when you've made 10 posts, so, really, it's all about how things go over the long term for any poster (as in, it's fine if someone makes 100 posts without receiving any merits at all, and then suddenly writes a banger that gets 10 merits [1]).

[1] The only issue around this, which I'm reluctant to get into because it's not important, is that some back-and-forth bouncing will sometimes occur when someone is near to a rank boundary.

---

It's clear to me (from some of the responses I've read) that I've done a poor job with explaining this idea. So, I'll take another shot at it, by way of analogy: Imagine that there's a service/website/app called Thoughts4Berries where you can submit very short pieces of original writing (as many and as often as you like) on any of a wide range of topics, and for each submission of yours that's deemed worthy (according to an algorithm that nobody seems to be able to fully figure out) you receive a prize of at least 10 berries (but you might also receive 50, 100, 250, 500, or even 1000+ berries if your submission is truly outstanding). Thoughts4Berries guarantees you that every submission received is always exposed to some chance of winning a prize (either now or in the future). So, basically, if you want berries, write something good and send it to Thoughts4Berries. You'll either get no berries for that attempt, or at least 10 berries for it. If you take the two possible outcomes and combine them with the previously-mentioned "greater-than-zero probability" guarantee, then you'll see that there's something very naive about this whole setup: the expected value is always positive. So, no matter what you submit, your berry balance will increase (not your actual balance, but the expected balance that your actual balance will tend toward over many attempts). Given the fact that the expected value of any submission is always positive, many people figure out that it's possible to farm berries by submitting lots and lots of attempts, but, if your submissions are terrible then the low chance of reward makes that a very slow process, so, a better strategy would be to submit many bad-but-not-terrible attempts. Of course, the best strategy (for everyone involved) would be to only make actually-worthwhile submissions, but, who has the time to do things sensibly, amirite? Now, let's say Thoughts4Berries comes to realize that most of what they get sent is low-effort junk that's barely worth reading. What can/should they do about that? There's a frustrated (and good-looking, too) dude working at Thoughts4Berries that goes by the name of, uh, MaxLegroom. MaxLegroom thinks that it's worth exploring the following idea: What if we extended an infinite line of "berry" credit to every user (to be offset against any prizes they might win), and then used that credit facility to charge 1 berry per submission? MaxLegroom sees that as an obvious way to reduce the amount of junk submissions by establishing a submission-wise "noise floor" that's too expensive for any rational user to keep operating entirely beneath. The only real question in ML's mind is where that cut-off should be placed (which is determined by the per-submission cost; maybe 1 berry is too much, he thinks).

OK. So. What does it look like when something like the above is applied to Bitcointalk's ranking system?

Instead of the "rank requirements" table looking like this:

Rank	Required activity	Required merit
Brand new	0	0
Newbie	1	0
Jr. Member	30	1
Member	60	10
Full Member	120	100
Sr. Member	240	250
Hero Member	480	500
Legendary	775-1030 (random)	1000

It would look like this:

Rank	Required activity	Required merit (less carry)
Brand new	0	0
Newbie	1	0
Jr. Member	30	1
Member	60	10
Full Member	120	100
Sr. Member	240	250
Hero Member	480	500
Legendary	775-1030 (random)	1000

Notice that I've changed nothing except for the description of the third column, because that's really all I'm proposing. I'm saying that the rank requirements should stay the same, but, instead of the third column referring to the required amount of merit, I think it would encourage better posting behavior if that column referred to the required amount of after-carry merit (where "carry" is something that slowly builds up as you post).

---

I do get what you're saying. The first issue I have with that approach is that it's subjective and therefore would be unevenly applied (different mods have different views concerning what makes something low-value; if I were a mod, for example, I'd probably leave many reports in an "unhandled" state because my anti-censorship inclination is much stronger than my dislike for junk posts). My second issue with that approach is that it applies pressure to a building block of the forum that's much more crucial than "ranking up": the forum's free-speech orientation should be preserved as much as possible, IMO, and placing it under unnecessary strain is unwise, I think. The right to post is paramount, the right to rank up... not so much. My third issue with that approach is that it's inefficient compared to a systemic disincentive that attenuates the creation of junk posts. (There's a software engineering idea that goes something like: "the fastest code is the code that's never executed", or, in the context of debugging: "the easiest code to debug is the code that isn't there". Applying that sentiment to moderation would go something like: "the easiest posts to moderate are the posts that were never written".)

I once tried to explain this whole "carry" concept in a PM, and I'll quote a small piece of that because it adds to the above:


Maybe. But, it doesn't make sense to me for the forum to "outsource" something so fundamental.

---

Let me express my interpretation of your point, as: "If the ranking-up mechanism is largely subjective, then I think it's a mistake for the ranking-down mechanism to be objective".

As in, it's unfair when the thing that lifts you up is hit-or-miss, but the thing that pushes you down always hits. I get that. It sounds very unfair. But, that conclusion depends on:

(1) How unreliable is the first thing?

(2) How strong is the second thing compared to the first thing?

To answer (1): I don't think the merit system is very unreliable. I can imagine that the whole apparatus must seem very unfair/rigged/cliquey to anyone that's struggling to earn merit, but, my own experience here has been that it's impossible not to get merited when you're actually trying to add value to conversations and you're avoiding conversations where you don't believe that you have anything valuable to say. Sure, maybe some of your really good posts will go unnoticed, but, if you keep hitting the "value" nail instead of the "quota" nail (or the "agenda" nail, or the "vendetta" nail, the list goes on), then, trust me, you'll stand out, and you'll eventually get enough merit that "activity" might become your ranking-up bottleneck (obviously, I'm not talking about you; you're already a stand-out member). I realize that I haven't answered any of your concerns around uneven merit distribution across the forum's different sections, but, I view that as an issue that would take me many posts to unpack, and one that's orthogonal to the adjustment I'm proposing.

To answer (2): I've purposely tried to make the ranking-down force weak compared to the ranking-up force (partially to address the perfectly-enforced vs. imperfectly-enforced mismatch). I don't believe it's possible to make the ranking-up force perfect without also making it unmeritocratic. But, I also don't believe that that means that the ranking-down force has to be imperfect, too. It just means that the ranking-down force has to be tuned to compensate.

---

There's more that I'd like to say, but, I'm in a pretty annoyed mood at the moment, and I don't think that a massive wall of text is anything that anyone wants to parse, anyway. (I've also noticed that I tend to write posts and PMs in a way where the things I've said in one part are comprehension-wise affected by the things I've said in earlier or later parts. As in, I often try to explain things conversationally, like I'm talking to a peer, rather than expositionally, like I'm talking to a student. So, the longer I make a post, or the more posts about something that I make, the more opportunity there is for unsatisfying discussion around not-meant-to-be-isolated parts of what I'm saying.)