Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Diseconomies of scale in fraud, spam, support, and moderation

danluu.com

Post reply on HN
Print view View original

31–40 of 216 posts

Re: Diseconomies of scale in fraud, spam, support, and moderation

#31

Post by Denzel »

crummy wrote: post #12
infogulch wrote: post #10

Way back in 2014 when Jeff Atwood (aka codinghorror) switched from Stack Overflow to creating Discourse, he gave a talk about it (see notes [1], sadly I can't find a live link to the recording anymore). He gave a pithy little explanation of why they built the Discourse trust levels system the way they did that stood out to me: > The only thing that scales with the community is the community. The point being, you have…

I wonder how that would work for Facebook, for example. Facebook Groups have admins, but for general posts to the public (or to friends) - could they be user-moderated?

We're essentially opining about the difference between governance systems.

Facebook (like most corporations) operates an autocracy. However, unlike most corporations Facebook generates 100B+ in revenue per year (more than most countries), is worth $1T+ (again, more than most countries), and home to 3B+ users daily lives (trifecta, more than all countries). It's time to acknowledge the company that walks like a duck and quacks like a duck, is a duck.

If Facebook was so inclined to change their governance model, a plethora of prior art exists, in public sector government implementations, for them to build from. An endeavor of this magnitude is neither easy nor simple nor palatable to more engineering-inclined minds, but it's perhaps generational opportunity for a person/company to truly lead and innovate on such a ubiquitous digital platform.

Re: Diseconomies of scale in fraud, spam, support, and moderation

#32

Post by CityOfThrowaway »

tareqak wrote: post #25
CityOfThrowaway wrote: post #20

The author seems to be mistaking the benefit of selection effects with competence. Those who work anti-fraud for platforms know that there is something like a market for fraud. There is a market clearing quantity of fraud that will be present on any platform, dictated by the gains from learning to execute the exploit, the effort necessary to do so, and the number of times one might be able to do it. Small platforms h…

OK, fraud exists. Irrespective of whether or not a platform is small or large, can they make their customer whole 100% of the time? Why or why not?

To take your question seriously – one of the primary surface areas for fraud is exploiting the customer service function.

Just to work one (real) example:

1. Buy expensive face cream from Amazon

2. Take out the expensive stuff and replace it with junk

3. Complain to Amazon that you got junk and need a refund.

The challenge here becomes obvious. Even in claims where the customers was allegedly a victim, some portion of those customers are fraudsters. So any policy that results in a large, profitable company will likely result in something less than 100% of legitimate fraud cases from being satisfactorily resolved.

In the case of the small business, they are very unlikely to be targeted by industrial scale fraud and will probably just be able to solve this with personal relationships and human judgment.

That said, the price you pay the small operator is very likely to be meaningfully higher than the price of the scaled operator.

If you care more about having near 100% probability of having fraud/abuse cases addressed than you care about benefiting from the economies of scale, then yeah, go with the small operator.

In things like communities, my general sense is that there is almost no beneficial economies of scale. So I like to hang out in smaller places. But in things like information retrieval (ie. Google), the benefits are usually worth it for me to use that.

Re: Diseconomies of scale in fraud, spam, support, and moderation

#33

Post by nonrandomstring »

ec109685 wrote: post #8

Amazon selling obviously fake SD cards / flash disks is baffling. It’s super easy to spot and fake disks cost users data, so you’d think they’d take that seriously. https://a.co/d/auXcxrZ ($29 for 1TB as their “overall pick”)

Some work I supervised sampling Amazon and Ebay products for vendor malware showed an absolutely shocking prevalence of dodgy goods.

I don't think they can stem the tide.

Re: Diseconomies of scale in fraud, spam, support, and moderation

#34

Post by Aurornis »

Denzel wrote: post #31
crummy wrote: post #12

Earlier quoted context omitted.

I wonder how that would work for Facebook, for example. Facebook Groups have admins, but for general posts to the public (or to friends) - could they be user-moderated?

We're essentially opining about the difference between governance systems. Facebook (like most corporations) operates an autocracy. However, unlike most corporations Facebook generates 100B+ in revenue per year (more than most countries), is worth $1T+ (again, more than most countries), and home to 3B+ users daily lives (trifecta, more than all countries). It's time to acknowledge the company that walks like a duck a…

> It's time to acknowledge the company that walks like a duck and quacks like a duck, is a duck.

Are you trying to imply that Facebook is a country? Because this doesn’t make any sense. Countries have an entirely different set of obligations, duties, and jobs to be done than a website dedicated to light communication and entertainment.

Re: Diseconomies of scale in fraud, spam, support, and moderation

#35

Post by kmeisthax »

>By comparison, on lobste.rs, I've never seen a scam like this and Peter Bhat Harkins, the head mod says that they've never had one that he knows of. On Mastodon, I think I might've seen one once in my feed, replies, or mentions.

I self-host Mastodon and my mentions are chock full of this one Japanese Discord cybercrime server[0]'s invites. This only started four days ago, but it's already making Mastodon almost as bad as Musk-era Xitter spam.

The front-line tool for moderating federated spam is to block the instance that sent it. A year ago, most spam were rando servers following everyone they could think of. However, the Japanese Discord guys are almost slightly kinda smart: they started registering accounts on smaller (non-single-user) instances. Because Mastodon doesn't support any kind of account verification[1], any instance with open registration (i.e. "just sign up") is effectively an open relay for spammers.

I don't consider this to be negating the post, but it is relevant context.

[0] They advertise DDoS, hacking, and stolen credit card services, or at least that's what I was able to glean from the Japanese language advert they sent

[1] This is why Google requires you to verify with an SMS code to do anything e-mail or YouTube related, for example - it lets them throttle how many registrations per year you can do with that number.

Re: Diseconomies of scale in fraud, spam, support, and moderation

#36

Post by kmeisthax »

nonrandomstring wrote: post #33
ec109685 wrote: post #8

Amazon selling obviously fake SD cards / flash disks is baffling. It’s super easy to spot and fake disks cost users data, so you’d think they’d take that seriously. https://a.co/d/auXcxrZ ($29 for 1TB as their “overall pick”)

Some work I supervised sampling Amazon and Ebay products for vendor malware showed an absolutely shocking prevalence of dodgy goods. I don't think they can stem the tide.

Amazon is perfectly willing to ban people from reselling their used Apple products. The only reason why they don't apply that treatment to memory devices is because Amazon runs on payola. Apple paid for market exclusivity and Samsung/Sandisk/etc didn't.

Re: Diseconomies of scale in fraud, spam, support, and moderation

#37

Post by miki123211 »

crummy wrote: post #12
infogulch wrote: post #10

Way back in 2014 when Jeff Atwood (aka codinghorror) switched from Stack Overflow to creating Discourse, he gave a talk about it (see notes [1], sadly I can't find a live link to the recording anymore). He gave a pithy little explanation of why they built the Discourse trust levels system the way they did that stood out to me: > The only thing that scales with the community is the community. The point being, you have…

I wonder how that would work for Facebook, for example. Facebook Groups have admins, but for general posts to the public (or to friends) - could they be user-moderated?

It would work for Facebook users, but not for Facebook the company.

This is essentially what Reddit does. It mostly works for communities about headphones, bicycles or water bottles, but not Donald Trump, piracy or child porn. Reddit had to close some subreddits down, not because their users complained, their users were perfectly happy and considered the moderation to be just fine, but the harm to society / the company's PR image / advertisers was too great.

Re: Diseconomies of scale in fraud, spam, support, and moderation

#38

Post by bjterry »

CityOfThrowaway wrote: post #20

The author seems to be mistaking the benefit of selection effects with competence. Those who work anti-fraud for platforms know that there is something like a market for fraud. There is a market clearing quantity of fraud that will be present on any platform, dictated by the gains from learning to execute the exploit, the effort necessary to do so, and the number of times one might be able to do it. Small platforms h…

That is not at all in conflict with the post. The selection effect is one of the sources of the empirical diseconomies of scale the author mentions.

Re: Diseconomies of scale in fraud, spam, support, and moderation

#39

Post by tareqak »

> In my social circles, many people have read James Scott's Seeing Like a State, which is subtitled How Certain Schemes to Improve the Human World Have Failed. A key concept from the book is "legibility", what a state can see, and how this distorts what states do. One could easily write a highly analogous book, Seeing like a Tech Company about what's illegible to companies that scale up, at least as companies are run today. A simple example of this is that, in many video games, including ones made by game studios that are part of a $3T company, it's easy to get someone suspended or banned by having a bunch of people report the account for bad behavior. What's legible to the game company is the rate of reports and what's not legible is the player's actual behavior (it could be legible, but the company chooses not to have enough people or skilled enough people examine actual behavior); and many people have reported similar bannings with social media companies. When it comes to things like anti-fraud systems, what's legible to the company tends to be fairly illegible to humans, even humans working on the anti-fraud systems themselves.

I think legibility is one aspect of it, but I think another could very well be empathy. If you want to support a user who has a problem, then you have to want to help them and follow through. If you want to make a customer whole after they lost money on your platform, then you want to pay them back and follow through. If you want to stop your platform to be a tool used to disseminate hate, then you want that outcome and move heaven and earth to make it happen. You have to want to see the problems and you have to want to follow through and fix them.

Re: Diseconomies of scale in fraud, spam, support, and moderation

#40

Post by CityOfThrowaway »

bjterry wrote: post #38
CityOfThrowaway wrote: post #20

The author seems to be mistaking the benefit of selection effects with competence. Those who work anti-fraud for platforms know that there is something like a market for fraud. There is a market clearing quantity of fraud that will be present on any platform, dictated by the gains from learning to execute the exploit, the effort necessary to do so, and the number of times one might be able to do it. Small platforms h…

That is not at all in conflict with the post. The selection effect is one of the sources of the empirical diseconomies of scale the author mentions.

My first read was that the author was claiming that smaller companies are at least as competent at moderation than big platforms.

In my first read, I took that to mean from a process and inputs viewpoint.

But reading it again, I suppose the author could be taking a more consequentialist point of view. They are at least as good (or much better), largely because they don't have to do much.

I suppose that's fair, but also a lot of words in the OP to state a truism :-)

Post reply on HN