I want to lay out why I think the warning-label model is probably insufficient for verified scammers, and why banning is likely the better policy. I'm going to argue from likelihoods and incentives rather than claim certainties.
A warning protects a buyer only if they see it, read it, understand it, and act on it before paying. Each of those is a step where some fraction of users statistically drop off, people who skim, people who pay first. So a warning does prevent some harm, but it doesn't prevent all of it, and we've seen cases here where someone was scammed despite a warning being present, because it wasn't read or was encountered too late. I'm not claiming warnings never work. I'm claiming they leak, that a percentage of users won't be protected by a label.
The more important point is about what a warning leaves intact. A scammer's effectiveness depends on accumulated (false) credibility markers: reviews, trade count, an aged account, badges, references. A warning sits next to that credibility but doesn't reliably remove it, and a crafty scammer can sometimes talk past the flag. Crucially, a warned account stays open; the warning is a sticker on a door that's still wide open, so it only works if the buyer happens to read it in time. A ban locks the door instead: the account can no longer operate at all, regardless of whether any given buyer is paying attention. So the warning reduces a scammer's success rate but leaves most of their persuasive power in place, while banning is the one action that plausibly resets that credibility to near zero and removes their ability to keep operating. Banning targets the thing that's weaponized for the actual mechanism of harm.
The platform already has reviews, ratings, trade counts, account age, and badges, all of which exist for one reason: because reputation affects how users behave. We built those features precisely because a buyer treats a 40-trade, well-reviewed, aged account differently than one created yesterday with no history. But if reputation affects behavior strongly enough to justify building an entire feedback system around it, then stripping a verified scammer of that accumulated reputation should be expected to reduce the harm they can do, even if the same person returns under a new name. You can't really hold both "reputation matters enough to warrant our whole rating system" and "removing a proven scammer's reputation accomplishes nothing", those two positions undercut each other.
Objection: "they'll just make a new account." This is true, and I'd argue it actually supports banning rather than undermining it. Re-registration isn't free to the scammer. A banned scammer has to rebuild from nothing, no reviews, no history. And a zero-history account is likely the weakest, most suspicious position a scammer can operate from, because their power partially comes from looking established. Assuming banning may not reduce how many scammers exist (they return), it still forces them back into the low-trust, low-effectiveness state each time, whereas a warning lets them keep operating from a relatively high-trust aged account despite the warning label.
There's a statistical difference in harm-per-scammer between "back to zero repeatedly" and "keeps the established-looking account." That difference isn't zero, and nearly every established marketplace or platform operates from the reasonably likely assumption that banning bad actors reduces harm more than the alternatives. So even under uncertainty, the expected-harm-reduction likely favors banning.
Even darknet markets, which operate in highly adversarial and reputation-dependent environments, generally ban confirmed scammers rather than merely labeling them. Market operators understand that bad actors can create new accounts, but they still use bans because destroying an established reputation, forfeiting account history, and forcing scammers to rebuild trust raises the cost of fraud. In other words, the goal is not to eliminate scammers entirely, but to make scamming more difficult and less profitable.
For what it's worth, a stricter "verified dishonest means removed" approach worked reasonably well in the LocalMonero era, which's one of the closest platform examples I can think of that achieved considerable success.
There's also a compounding effect. Under a warning-heavy model (some edge cases get removed AFAIK), every scammer we leave active has a cost to the total # of scammers going around. The platform accumulates flagged-but-around bad actors, each one an aged, somewhat-trusted-looking account that someone may not vet carefully. A growing field of labeled-but-live hazards, like a backyard with mines that have only flags on top of them, that we've seen some users verifiably walk over despite the warnings. I don't have analytics to quantify how much this degrades median trust over time, but I believe the effect is detrimental. Granted, this is purely anecdotal and I'm not claiming it as hard evidence, but I've talked with a number of people about the platform, both inside and outside, and a not-insignificant number report that they think there are too many scammers around here. A marketplace's reputation is one of its more fragile assets. To be fair, this is a Monero platform and the situation is not entirely escapable.
On evidence: there's research on ban-evasion in online platforms suggesting that banning tends to degrade fraud success even when offenders return, the general finding being that reputation and accumulated credibility don't transfer cleanly to a new account. This's suggestive rather than conclusive, some of it is preprint or industry research, and platform dynamics vary. But it's at least consistent with the incentive logic above: removing the account removes some advantages, while flagging it has a different set of effects.
I'm also referencing some related writing of
AshleyJones here:
> "I'm a proponent of freedom of speech too, obviously. However, imagine if a guy went to a movie theater, bought a ticket, and then in the middle of the movie got up in front of the screen and started reciting poetry in as loud a voice as he could muster. If someone did this, they'd be kicked out of the movie theater, and rightfully so. Nobody is trying to 'censor' them or stifle their freedom of speech -- but the movie theater is selling a product (an enjoyable movie-watching experience) and this person, who entered the theater voluntarily, is acting in bad faith and ruining the product for everyone else. That's a silly scenario, but this is effectively what people who act in bad faith in any kind of online community are doing. As the manager of a movie theater, or as an administrator of a website, my opinion is that you not only have a right to remove these people, but an obligation to do so. A 'mod log' adds transparency, so that everyone else on the website can understand exactly why certain administrative actions were taken. Without this transparency, the guy in the example before could go around town telling everyone that the movie theater kicked him out 'for no reason,' when that's obviously not the case. A transparent mod log gives a community more trust in the administration, because if they DO happen to see an action taken on the mod log that they disagree with, they can voice their opinion then. In 99% of scenarios (deleting spam, banning obvious bad-faith users) the community will thank you for dealing with the problem."
Source:
https://dukenukemis.cool/textfile95.txtMy understanding of how things work around here is based on my own observation, and I fully accept I may be missing context or be wrong about some of it. If you see something I've gotten wrong, please correct me. I'm sharing this as my own view to get the community's opinions and feedback, since I think it's an important topic worth discussing further.
Tip Monero to SoulReaver 
8AJJt5FoxgKapYk69YgSLXPaSBYd88hghg32ie25ZXGSFFYgVeKT4ytLkPoCsnJGE3gQninbK6JDUAX4qNCRUqfXFwXe3Hm
Publish Tip to SoulReaver 
8B9ZmojrPx5c19RvWwYkYkHfgX9Cu29av6vpnWPF61jwdrs6ujRcrUaUrS8Q9UPs4EhwN7ibwJGRmYMKwWkfgwPM8ngdgXL