Back in the days when trust mattered...Back in the days when trust mattered...
Recent ranks posts by time. top ranks the posts by most sats for a specified time bucket (day, week, month, year). But hot is different.
hot used to be something I didn't understand. Posts on the SN landing page weren't just ranked by number of sats, but by some complex combination of sats zapped by "trusted" users. And trust was unclear to me: SN didn't display a trust score. It was something you had to guess at. If you used SN very much, it was clear that trust on SN was gained by zapping things that lots of other stackers would later zap (I'm probably still getting this wrong -- as I said, I never really figured out trust...trusting instead @Undisciplined's call to zap more)
No Trust November is all about the sats!No Trust November is all about the sats!
Since #1362477, hot is ranked the number of sats divided in half for every four hours the post has existed.
(That sounded a lot simpler in my head than when I wrote it.)
The point is, if you want to see your post (or anyone else's) on the front page, it should be easier to figure out how to make that happen.
Specifically, you just need to zap it. It's as simple as that. New account, old account, power user, or anon -- all zaps count the same. It's just a market where real estate on the home page is auctioned off to those willing to buy it.
This makes me think that perhaps ranking isn't the right word at all. It's more like an auction for the real estate on SN's home page, where the top of the page goes to the highest bidder.
What makes this even cooler is that it's a kind of collective auction -- all the stackers who zap a post contribute to the "price" it's willing to pay to be on the front page.
All zaps count the same except downzapsAll zaps count the same except downzaps
The main reason SN has a 30% sybil fee on zaps is to prevent someone from gaming the system by zapping themselves with alternate accounts. If 100% of a zap went to the poster, that poster could zap themselves with no cost, making sats a kinda useless signal. If 30% of each zap goes into the rewards pool, it makes it more costly to do this: you have to spend at least 3 sats to buy 10 sats worth of home page real estate.
This is accounted for in how sats count towards a post's bid for home page space:
((sats + boost + (comment_sats * 0.25)) * 0.3) - downsats - (comment_downsats * 0.1)
As you can see, zaps, boosts, and zaps on comments only count as much as their sybil fee (30%). Whereas downzaps count 100%, because 100% of downzaps go to rewards.
This means downzapped sats are roughly 3x as powerful as regular zaps. So if you really want to feel powerful, do some downzapping.
Should boosts be as strong as downzaps?Should boosts be as strong as downzaps?
You can also see that boosts currently only count at 30%, even though 100% of a boost goes to the reward pool (making boosts much harder to game than zaps).
@k00b pointed this out and that it might make even more sense if the full number of sats in a boost counted toward a post's bid for space on the home page.
TIL about the purpose of the 30% Sybil fee!
So, what advice would you give someone who is sharing quality content but is virtually ignored because they A) can't afford to boost, B) has integrity so they don't self-zap, and C) shares content across all territories so the trust score isn't high (although I'm not really exactly sure how that works..?)
I think people do appreciate it when it's obvious that someone spent some time and thinking on their post or comment. They don't always land (I spent a week writing a post about Frankenstein and bitcoin and that sure was a dud).
my advice would be to find a niche that interests you and then be consistent. Post about it on some regular schedule. Do some research on it, try to post something about your niche that maybe isn't easy to find or isn't obvious. if it's something you are super interested in, eventually others might catch your enthusiasm.
It sounds like you are more left leaning. I'd be very curious to read quality posts about immigration topics or about climate change. Those are under-covered here and although there are many people who might react somewhat harshly, there are also a number who would appreciate thoughtful posts on those topics. I'm sure there are others.
I don't think people are that harsh on left leaning views as long as they're presented respectfully and with a willingness to discuss, not presented dogmatically and shrilly (though for some reason progressives have a hard time with the last part)
I'm pretty open about my status as a commie and a statist, at least compared to the average SN denizen
Boost is self-importance, so I don't think so. But perhaps zaps and downzaps should be equal
The problem with shoulds when we aren't using trust is that the proof-of-sacrifice differs among these things.
If we ignore territory revenue (and conflicts of territory founders) by limiting proof-of-sacrifice to reward pool contributions:
We optimized the current rules for clarity. If we want the system to be free of shoulds though, the above is how I'd arrange things.
If we make zapping more powerful than boost, then I will sybil-zap because I get more power for less sacrifice.
I do really like the clarity. I also feel like it's not really getting gamed.
right now, I would say a downzap demonstrates more signal than a zap. Since SN is a pseudonymous site, it's hard to know if anyone is self zapping (collecting 70% of the value of their zap). In order to get the benefit of a 100 sat zap, a self-zapper only has to spend 30 sats (and take the trouble of setting up an alt account).
However, a 100 sat downzap costs 100 sats because the downzapper cannot recoup sats by self downzapping. So that's why I'd say downzapping is a stronger signal than zapping.
Boosting is a self-zap where we can see that the self-zapper didn't receive any portion[1] of the sats back. In order to get the benefit of a 100 sat boost, they actually have to spend 100 sats.
A boost may be a sign of self-importance or ego, but it may also be a way of saying "I think this post is so worth your time that I'm willing to spend to get you to see it." A downzap may be a sign of spite or malice or it may be a selfless act of community policing, like picking up litter so other people don't have to see it.
Rewards make this messy because if some of the boost goes to rewards that the boost-payer receives, they do receive back some of the sats. But it's not as sure a thing, I don't think. ↩
Thanks! This makes sense. The problem really is the sybil-zapping in this and I'm looking at it from a too naive perspective.
I agree. Downzapping doesn't provide benefits if the goal is to earn sats. That's why it's good - it serves to filter bad content from SN. The benefit is long-term for those who consume it. It's altruistic.
If this works like you say the “system” is easy to fool. This would mean that many interactions here are artificial from someone that learned to manipulate it. But isn’t what I saw. I thought the trust works using many variables, but I didn’t understand at all too.
To have a ranking, we would need another form of evaluation, maybe one per user, like an upvote? As you mentioned, the "sat by sat" is a market for whoever spends the most, not for the hot, as the name suggests.
I'm waiting for the adjustments to give more visibility to old posts.
This seems like not quite the right implementation of a hot feed. The half-life part makes sense, but shouldn't it be based on when the zaps came in rather than when the post was created?
If a post blows up three days after it was made, it seems like it should show up in our hot feeds to let us know something's happening that we may want to pay attention to.
It would be cool to have some sort of trust toggle for each of these rankings that basically flipped them between global and personal ordering.
I've explored this some. Ranking works on item age and sats. To let new zaps on old items pull an item up in ranking, we'd replace item age with an
effective_agethat's influenced by the recency of zaps. Computingeffective_ageis where it gets tricky but I think I can figure it out.i is the item and j is the zaps on item i
σ controls the half life
Probably just drop the zaps that are older than some preset amount too
we can weight it by the exchange rate at the time ~lol
NONO
Because then the units of the rank would be in dollars not sats!
~lol
the exponential decay pretty much zeros out old zaps contributions anyway, so the only different it'd make would be the computational speed of filtering out old zaps vs. extra calculations by including them
Computation is what I was thinking about, but I didn't think about the computation of filtering
Nice! That's roughly what I arrived at as an accumulator for making a resurface decision. When it exceeds a dynamic threshold, I'd move forward the
effective_age.I've been a little hesitant to give full control to recent zaps alone, but I'm probably just FUDing myself and it's worth experimenting with.
Edit: It also hadn't occurred to me to use it raw like this until your comment. I think this would make it worth being in the back catalog.
I think the difference would be pretty marginal. 95% of zaps happen in the first 48 hours. At least this gives a chance of good old posts to resurface.
We'll see. I think if posts are able to resurface and people can earn rewards for zapping them when they do, then we'll see more zaps on older posts.
(assuming I described how it works accurately) seems like it could lead to a way to deal with evergreen content, too. There might be a risk that the
hotfeed becomes less lively.Always one of my main concerns
My question about the trust score isn’t about Trust November, it’s about crossposting. We know that each stacker has a trust score per territory, so my question is: how does a post/comment get ranked when it belongs to more than one territory?
Great question. I would expect it to be treated separately by territory when you are specifically viewing posts in a territory but averaged with some weighting when you're viewing site-wide posts.
Currently, there is no trust. So, when you view the home page, you see all posts (minus any things you have muted) ranked by sats (which become half as "valuable" every 4 hours).
When you click in to a territory, you see all posts in that territory sorted by sats with the same time decay.
I took the question as how trust would influence the rankings if we had trust based rankings again.
I ignored trust per territory when I implemented crossposting because it wasn't relevant. If we introduce trust again, that's something I'll figure out then.
That's a big if
I'd expect trust to matter more at scale but so far I don't feel like it's a major loss.
From what I can tell, based on an eyeball test, things seem pretty healthy.
At scale, I suspect incentivizing downzaps would be most of what's required.
Scale is also usually a FUD-y thing. Evoking it sounds smarter than it is. haha, when the complexity of the system explodes and is relatively unpredictable, I predict xyz
Fair enough. I'm not even thinking about complexity exploding, though. Any improvement to what's relevant to me matters more when there's more stuff to choose from.
I think the trust score should matter more or less depending on the number of active users.
From a reader's perspective, it seems like it matters more specifically when there are more posts, regardless of how many users there are.
ah, that makes sense.
Daily rewards are still based on the trust score.
It’s a possibility, maybe @k00b can clarify. The same question applies to earning trust score on crossposted posts/comments. It looks like we might be gaining trust score in multiple territories.
For example: a post/comment crossposted to two territories. If we receive a 100 zap, do we gain trust score in both territories? 100 | 100, or 50 | 50, or something else?
#1406788
Just to be clear, we’re not earning trust score?
you are earning or losing trust in all the crossposted territories currently, but trust has no impact on ranking.
That was exactly my question, how are we gaining or losing trust score in each territory? Is there some kind of weighted formula behind it?
it's the same as before but now in all of the territories
A lot more sports posts appearing on the hot list since trust element was removed. I never really understood how it worked before that stackers posting in sports had high trust scores within sports but not more broadly on SN.
Anyways I prefer this format better. It seems to offer more variety on the hot page with content from many different territories.
I do miss the old format for rewards payouts though. I used to get a lot more rewards. Haha
I've really enjoyed seeing the home page get a shake up!
Not in line with my preferences, but who am I to gainsay the vox populi?
Two reasons you’re getting fewer rewards: the daily incentive dropped to 10k, and I think now you get the territory fees instantly, whereas before you got them with the daily rewards.
Yes I factored in the territory fee difference. You’re probably right that it is mostly due to the decrease in daily incentive.
interesting. i posted something today and it was pretty high on the frontpage and now slowly goes down. what i didn't think about was the timing of me posting the link. since i am from switzerland and i posted the link in my morning, people from the us might never see it on the frontpage. if it would be important to me how that link ranks (it does not in this case), i would need to take the timing into account. also i am just assuming that the us has the bigger audience than the eu, but maybe that assumption is even wrong...
At the moment, SN seems to be more active during US daytime hours, so a post in the morning hours of the EU tends to drop down the home page by the time SN hits its most active.
But this also means that EU morning hours have less competition for real estate.
good point. as so often in live, there are pro and cons
I had
wronglyassumed boosts went to territories and rewards. Isn't that how it used to work?Edit: seems i was not wrong?
boosts do go to territories and rewards
Wow! Does that mean territory owners can boost their posts with a 70% discount?
Always has been.
Boost = self zap (for ranking) + territory subsidy
big reason i lowered my fees. bring the boosts!
This gives a lot of power and responsibility to territory owners. Maybe too much! Shouldn’t it be the other way around? Like 30% for the territory and 70% for the rewards?
They pay dearly for the power relative to others
indeed
Thanks for clarifying!
Yeah, I"m pretty sure I got that detail wrong.
Love this explanation! Makes the whole ‘hot’ system feel way less mysterious turns the front page into a kind of community auction, which is a really cool way to think about it. Downzaps being stronger than zaps is wild but makes total sense.
Fascinating shift! From opaque trust to a clear sats auction zaps now literally buy your spot in the spotlight.
shallow
I've been using the downzap feature since it was first implemented. Especially against the right wing propaganda spewed on here.