Amazon Froze My Account and I Still Don’t Know Why | Hacker News Reader

Amazon Froze My Account and I Still Don’t Know Why

bardsandsages.com

235 pointsilamont125 comments

I think this is the future rise of the robots. It won't be Skynet, with clearly malicious bots trying to destroy humanity. It will be capricious and arbitrary bots, locking us out of our bank accounts and merchant accounts with little or no recompense. We've already seen it with Google, now here with Amazon. How soon until your phone gets turned off, and that's just the end?

And if you're in the US, there are no restrictions on corporations restricting your rights. Only the government. The more that is privatized, the less rights you have.

>And if you're in the US, there are no restrictions on corporations restricting your rights.

That's not precisely true, though in practice it may depend on the willingness of regulators to enforce existing rules, or people being willing and able to sue in court. Our corporate overlords can often count on those factors working in their favor.

> the willingness of regulators to enforce existing rules, or people being willing and able to sue in court...

Regulatory capture and mandatory arbitration agreements buried in the click-wrap.

Our "corporate overlords" have structured those factors to work in their favor.

In the past we spread our patronage and investments around to protect against the failure of businesses.

Looks like in the future we'll do so to protect ourselves against AI.

Maybe there will be a flipside to this. Perhaps we can hope for the ability to have our own AIs that work on our behalf.

Maybe if the story author had written a bot to reply "I don't understand. Please tell me what I did wrong." to every bot-generated email from Amazon, he could have made a more entertaining story. He could include some graphs of the reply-counts, estimate the AZ bot's latency, and draw out some maps of the various routes taken by the messages.

So, just like the government, then.

Except that with many governments, you have democratic or procedural recourse (When some low level clerk at the DMV arbitrarily cancels your driver's license, you can almost certainly escalate, demand more transparency, etc. When some low level clerk at Amazon closes your account, you are screwed. Appeals will give you no transparency.)

With business, your only recourse is the equivalent of moving to a different country.

It's the perfect storm of bureaucracy, the-computer-said-it-so-it-must-be-true, for-profit motives, and lack of transparency.

Yes, when the government is acting as intended, this is the case, but it's far too easy for evil people to use government power to do evil things. And the more powerful we make government, the more this happens. And in those cases, you seldom have recourse.

If a company routinely closes its customers' accounts when there's no good reason to won't it open a space for a competitor who is known for not doing that? Seems like it's hard to have an equilibrium where one can't keep a bank account open and no bank is trying to grab more market share by reforming its system to prevent unnecessary closures.

Maybe. Google has economies of scale on its side already and automation allows for higher profit, so so long as it isn't regulated it's practically and nearly impossible, an uphill battle.

How often does a large company really get brought down by screwing over some small subset of their customers?

Guess I was more responding to the speculation of a Skynet scenario than of a company screwing over a "small subset" of customers which plainly does routinely happen.

It's not as if the market can magic gigantic competitive companies into existence. Amazon is a monopoly because of its huge scale. It happens - almost incidentally - to have an effective monopoly on a large sector of the self-publishing market. But that's a small part of a much larger online retail marketplace monopoly.

This means it can dictate terms to authors, and generally do whatever the hell it wants to. There is nowhere else that authors can go without losing significant sales.

How easy do you think it would be for a competitor to build a monopoly the size of Kindle from a cold start? How much cash and other resources would be required, and how long would it take even if the cash and other resources were available?

And after all that - how profitable would that competitor be compared to Amazon, and how likely to provide a reasonable return on IPO?

It sounds like we'll need government agencies, even perhaps tying it into a local police forces' work, to take these issues seriously - as serious as a crime - to make sure answers and reasons are found. Ignorance is bliss and reduces cost for a for-profit business, however it's harmful and negligent and perhaps verges on abusive.

Not a law enforcement problem at all, unless there's some kind of fraud going on. Nobody has a right to an Amazon account. That's totally up to Amazon to manage how they want to.

Perhaps not currently a law enforcement problem - the police are to serve and protect society. We don't let private companies do whatever they want, and this may evolve into a situation - especially if using AI is incentivized by cost benefit - where police can be further utilized.

Maybe the structure of it will be that we have a right to our data and Amazon is obligated to provide access to it unless a reason is provided.

Its all fun and games until someone dies in an elevator...

https://www.youtube.com/watch?v=sAz_UvnUeuU

For now, yes. Historically, corporations become more and more abusive until there's a public backlash that gets them regulated. Something like "there must be some clearly-defined way to get a response from a human" seems like plausible legislation.

Why do you need a law like that when the person can vote with their wallet?

In a smaller town, "voting with your wallet" is just choosing between Wal-Mart and Amazon. Wal-Mart for many is just a bad, but for different reasons (well, not that different). Either way, you're only getting a choice between the two big companies that are killing smaller retailers.

Eventually, politicians that like to (pretend to) champion small business (but then write law helping giant corporations at the detriment of small business) will have to put their money where their mouth is.

What if you're in a small town and the only grocery store shuts down because it's not profitable. Should the gov't force them to stay open?

Nobody has a right to an Amazon account.

Well, yes. But when Amazon are realistically the only place to sell ebooks you have to balance the right of Amazon to deny an account to whomever they want with the right of an individual to sell ebooks. With the supermonopolies of Amazon, Google, Facebook and Twitter these battles will be fought in the courts sooner or later.

We've been through this with music/movies. If there is no legal avenue people will pirate.

If you can't buy The Return Of The War Of Thrones on Amazon you can pirate it. If you can't sell your book, piracy doesn't help you.

Maybe, but what if it's your bank account? Locking you out of your funds can effectively mean the end of your life in short order.

There are strict regulations regarding bank accounts.

Perhaps there should be strict regulations about merchant accounts.

This isn't some tiny startup selling beanie babies here. This is the opaque policy of a multi-billion international, that is suffocating small business.

If by Merchant Account you are referring to the financial account that holds funds owed to merchants, I agree. There are regulations that cover this, but they arn't as strict as those regulating consumer banking. I'm not quite sure what a reinforced regulatory framework for this space would look like, though- the devil is always in the details.

Monopolies need oversight.

When there's no significant market oversight - as there isn't with a monopoly, by definition - governments need to step in to force reasonable behaviour.

No one has a right to credit, but we have strong regulations around credit scores because they determine so much about our lives.

Actually any company provides accounts should have a time-frame where reason of closure or disablement should be given to the account holder. That is where the laws need to change.

So less like 1984 and more like the movie Brazil? It was not too long ago that Amazon used to be known for amazing, human, customer service.

Great customer service but ruthless with their suppliers. The author here falls into the latter camp.

Good point, though I feel like their customer service has slipped too - e.g. allowing counterfeits to mix into supply of products is not customer friendly.

The consequence is that you have to start thinking of your entire digital life as disposable. I hope a company will rise to solve this problem and fill this need.

Interesting idea, how does one do this?

Make friends locally, save a ton of money (in case you can't work virtually later on), have entertainment locally (backed up or IRL), what else?

Yes, yes, and yes. And local backups of all the things. Don’t rely on anything you don’t host yourself, on your own metal.

With the exception of a VPS and possibly my AppleID, I could probably lose all my online accounts today and it would not result in meaningful data loss.

I’m planning to move my (VPS-hosted) Email on-site one day, and after that’s done my ISP is the last remaining point of failure.

> your own metal

It's absolutely maddening that I can't update my video card driver without logging into an nvidia account. That's the direction "your own metal" is going to keep going though.

If you roll back your driver to an older version of GeForce Experience - one that doesn't require the nvidia account - wouldn't that let you pull down just driver updates?

eg as long as you don't update the GeForce Experience program, you should be ok?

Exactly, you almost need to be sort of a prepper, with some cash on hand. But I was thinking of the whole vertical stack (backup email address, cell phone, credit cards). I’m thinking “continuity of business” preparations.

I think the first important step is to control the domain that your email address is on.

True, but my point is with your digital identity there is always a 3rd party risk. You might want to pay a reputable email provider for your primary email and pay a second reputable email provider for your secondary

If you own the domain name and back up your emails regularly I would say having the provider shut down wouldn't be more than a nuisance.

I think the market leader here is Synology, but I haven’t carefully checked out their competitors, and it is only a subset of what you want.

The TV series "Little Britain" had a regular sketch featuring an apathetic bank employee who just checked the computer after every request made by a customer and read out "computer says no" every time, never with any justification to the seemingly completely arbitrary rejections.

For banks it reflected the reality at the time (2000-2007), but it's creeping into more and more areas.

> For banks it reflected the reality at the time (2000-2007)

Today, too. Just try to get a mortgage at Chase.

In later episodes, the character turned not just apathetic but hostile and sometimes didn't bother even checking the computer before saying "computer says no".

As other commenters pointed out elsewhere in the thread, bureaucracy is a way for people/institutions to say "no" and automation is just a modern layer on top of that.

I feel like the "rise of the machines" won't be some bloodbath with terminators or AI killing people outright, it will just be that everything will be a Kafka-esque maze of bureaucratic complexity with no human recourse when you get caught in the system. It's like those Coke Freestyle machines, where you get stuck in line behind someone trying to figure out the touchscreen system when all you want to do is get a Coke. The future of society is all of us frustratedly waving our hands in front of an automated paper towel dispenser before we just give up and wipe our hands on our pants.

And yet, it's not technology that does this. It's the inherent nature of systems, and has been around for a long time:

https://www.youtube.com/watch?v=JtEkUmYecnk

At times it seems the whole point of bureaucracy is to empower people to say "no" even in cases where the complainant has a valid case. You save money by nickles and dimes by forcing people to be a slave to the process and not allow any pesky sense of justice or empathy to impact their moneymaking performance.

This is already the case. Much of the adult condition consists of trying to prove one's case to bureaucracies: why you shouldn't have incurred that overdraft, why you shouldn't have been denied that insurance claim, why you think there's a mistake on your bill -- and merely getting a human on the phone isn't sufficient to resolve your issue. You should also come armed with ample documentation (which you may not even have) to support your case, and be prepared to set aside a chunk of time as you bounce from department to department. This is part of why money-poor and time-constrained people have such terrible outcomes: each mistake hurts more in proportion, they lack the funds to hire an expert (e.g. tax professional, lawyer), and they're constrained on time to deal with offices that are only open for less than a quarter of the week.

There's a march towards more automation, more externally-opaque algorithms, and less humans to escalate to, but the core of the problem is quite old. Instead of a polite-if-flustered customer service representative telling you they don't know, you need to speak to someone else, you'll have to scrape the secret number from a 3-year-old reddit post or publicly shame them on Twitter to get answers, but the outcome's the same: the better-equipped and loudest voices win, while there's a bunch of others who get shafted with no recourse. Eventually, you quietly learn to keep thorough records, emergency savings, and eat the loss, because unless you can fight these sorts of issues in a higher forum where damages are an option, fighting them more than two roundtrips deep isn't worth your time.

I've spent a lot of time trying to come up with a business case for this problem. Something like : 'For $x I will call your bank/real estate agent/local council/supplier and sort out $problem. If I don't solve it, you don't pay'. Somewhere between a lawyer and a PA.

But I couldn't get around the privacy implications - I'd need all your passwords, birth dates etc and would need to either certify myself as your agent, or pretend to be you.

Edit: dizzystar and other below actually do just this thing for dealing with Amazon

Reminds me of the classic tool track featuring 10 minutes of automated phone system hell: https://www.youtube.com/watch?v=4cem5PEF9Bs

One of the things that frightens me most is Google Hire. What happens if for some reason it filters you out from everything? Kafka-esque can permanently ruin your life.

Imagine the multi-dimensional cross-checked ML driven future of credit reporting. You go in to apply for a home loan... "We see here that you missed a payment on your car loan in January, yet we also see from your trending posts that during that month you spent an estimated $408.34 on non-essential expenses during the same period based on your mapped locations. I'm afraid we cannot offer you anything at this time."

Or the flip side where you benefit because the system understands your actual risk profile rather than being at the mercy of highly error-prone credit reporting. I had 5 tax liens on my credit from the state of Missouri, despite never living there or operating a business there. It took me a year and a lawsuit against Experian before that was fixed. And this is despite having documents from the Missouri courts discharging the liens and certifying They were imposed in error.

Sounds like a visit to every French government office to which I’ve ever been subjected to.

This is just the first step, it will get worst.

I'm at still irrationally frustrated about my fledgling MailChimp account being flagged and locked months ago for no discernible reason and with no meaningful option for recourse. I am now a happy MailerLite customer and will not use or recommend MailChimp.

Sounds more like you are rationally frustrated about losing your MailChimp account, to me.

As many I'm afraid of a future (and present) where big platforms like amazon, google, facebook are essentially without alternative, and laws governing competition and thelike become a farce in the face of "rule by terms and conditions".

In Germany though, we recently had a nice case about a nation-wide football (soccer...) arena ban, in which the constitutional court ruled that this constituted a significant obstacle to taking part in regular social life. They pretty explicitely stated that this is not limited to football arenas, but can in principle apply to facebook and thelike. According to this ruling, they'd have to come up with a good reason for banning a user and there would be potential recourse. Very encouraging, really, since the ruling just admits what many already claim: the big platforms are an essential part of private/professional life, really they're more infrastructure than a choice made by the user.

Having had to work with Amazon support in a variety of capacities for many different businesses this is a common response to Amazon's support.

Contrary to what a lot of people think these are generally not robots. They're just people with a very short training period governed by an extremely didactic decision tree. If you do not give the support person what they expect using Amazon's unique definitions for terms which are often orthogonal to the dictionary definitions (e.g. you are supposed to get IP owners to 'retract' complaints even if you are really supposed to 'resolve' those complaints and there is a giant distinction between 'counterfeit' and 'inauthentic' products in Amazon-land) .

This often causes a lot of frustration as people shake their fists at the injustice of a highly bureaucratic system. Fist-shaking doesn't really do much and it just sorta amps up the frustration level. Sending lots of messages in also tends to befuddle the support people and/or cause giant problems if the wrong thing is said.

In practice what it means is that people who try to sell on the marketplace without knowledge of the opaque rules that govern it will tend to get weeded out. People tend to expect that the systems that they rely on are governed by moral rules about who is right and who is wrong. But it is really more easily navigable by people who are savvy.

What's fair doesn't really enter into the picture and the way that Amazon's systems in particular work tend to run against the average American's expectations for justice (expecting to be innocent until proven guilty, that laws and procedures be readily comprehensible, that judgment be even-handed and based on precedent, and so on).

However unlike some other big tech cos like Google/FB you at least have avenues of appeal beyond suing them. If you have a business problem with Google you will probably have to do something insane to get their attention.

> In practice what it means is that people who try to sell on the marketplace without knowledge of the opaque rules that govern it will tend to get weeded out. People tend to expect that the systems that they rely on are governed by moral rules about who is right and who is wrong. But it is really more easily navigable by people who are savvy.

How does one avoid getting weeded out and become savvy if the rules are opaque?

The rules aren't opaque. There just isn't any ambiguity to them, which means that everything is exactly written with no room for Interpretation.

I explain Amazon support like this: if you told someone to grab money from the bank, you haven't specified the instructions, so they will claim they have no idea how to do it.

If you tell them to drive to the bank, walk in, go to the teller, not the ATM...

This very specific instructions will get the response you need.

With experience, you start to get a hang of the lingo. It does feel like you are talking to bots, but you are really talking to humans who click the next step in the Q&A. They probably could hold your hand, but they don't appear to be trained to do so.

It's highly unlikely that you'll get shut down full stop right off the bat. You have a lot of room for error, but if you get frustrated quickly, you can get blocked at some step, but really that's on you.

The situation in the article would have taken me about 1 day to resolve, but the key takeaway is the books were up for sale again because the author did exactly as the instructions said.

'Opaque' as in 'generally opaque from the outside.' You learn in the process that you're describing.

Hilariously, it's to some extent like playing Zork or a similar text-based adventure game but with real money on the line.

The easiest approach is to just ask the rep to install Teamviewer, it will help make the process much less opaque.

https://imgur.com/a/yaI4B

The books are up for sale again for now, but the seller has no way to know whether this will happen again or what will happen to their account if it does. That's what opaque means.

Would you be able to resolve this in such a way that you'd know nothing bad was going to happen to your account as a result of your resolution? Would you be able to do that just with information that Amazon makes publicly available?

In the Amazon seller world, there are companies dedicated to dealing with Amazon and navigating through all the different warnings, suspension emails, and plan of actions.

For example, I'm currently dealing with a counterfeit policy warning for a product I've never sold or heard of... If I can't figure out the right words to respond to this email, I will have to fork up $100-200 to a consulting company so I don't put my Amazon account and business at risk of suspension- just so I can learn how to tell Amazon they made a mistake.

What serious consulting company in the amazon seller space charges less than 1k to do anything? I have yet to find any, and reinstatement handling usually starts at 2-2500, sometimes more for the ones run by lawyers or high-level former amazonians

It really depends on what you need. I'll send you an e-mail.

Oh yes. Knowing how to navigate the maze of Amazon support is extremely valuable to sellers. I sometimes do consulting for them to help with this very issue.

Support is exactly what you ask for, nothing more and nothing less. It also helps to know that whatever country they use has bans on things we in the US take for granted.

>Contrary to what a lot of people think these are generally not robots. They're just people with a very short training period governed by an extremely didactic decision tree.

That's essentially a bot made of flesh and bone. If there's no room for critical thought, you are essentially just software that breathes.

> If there's no room for critical thought, you are essentially just software that breathes.

"Today, Amazon is thrilled to announce: Mechanical Turk On-Premises."

people who try to sell on the marketplace without knowledge of the opaque rules that govern it will tend to get weeded out.

This person did understand the rules, and followed them quite carefully.

Also, are you sure that automated support isn't being used in some departments? Human reps are a huge cost, and considering Amazon is investing so heavily in AI/ML, it would seem that its own customer communication channels would be a natural place to apply the tech.

There's some automated stuff for routing but generally speaking you will be answered by a person. They may get blocs of text as prompts to use and the kinds of responses they can give are quite restricted by their policies but they are still real people.

Also in this case she thought she was compliant with the rules, but according to Amazon (which is not always reasonable) she was not compliant. Amazon support tends to send out these kinds of robotic emails that require a very specific phrasing in response (again, like I said elsewhere in this thread, similar to playing a 1980s/90s text adventure game).

If you don't get the adventure game reference, in those kinds of games you would have to solve bizarre puzzles using a very specific syntax. If you had to open a lock with a key, you might have had to try a bunch of strings like >"use key in lock" > "You cannot do that." >"turn key in lock" >"You cannot do that." >"use lock with key" >"The lock opens. You can examine the drawer."

Once she included the magic phrasing that Amazon requested, her account was unfrozen.

You may think that this is just because she is some tiny author, but you would be mistaken. Even top ten US/international sellers with tens of millions of dollars of sales per year are held to the same silly standard and hitched up by robotic support people even when they pay over six figures for special reps at Amazon.

>But it is really more easily navigable by people who are savvy.

Not really. I know plenty of 7-8 figure sellers that have the exact same issues all day long.

Hell, pharmapacks, biggest amazon seller by number of feedback, publicly talked about the same issues in their inc profile 2 years ago https://www.inc.com/magazine/201603/burt-helm/pharmapacks-am...

>But Vagenas's desk is constantly cluttered with products that have caused problems. Each day, the Seller Support group takes down one or more Pharmapacks listings without warning because of customer complaints.

They're savvy, yes, but they have a ton of issues just like every other big seller.

I wonder if there is a list of buzzwords you can send to the autoreply that will get it in front of an actual human. Like:

"Hey, seems like your bot is broken and I need to talk to a human, so I've copy pasted the following terms to attempt to elevate the issue and reach a human:

emergency fraud lawsuit hacked liability jeff bezoz better business bureau fuck shit escalate critical severe engineer SLA reliability false positive PARSE ERROR..."

A corporate version of the old NSA salute. al-qaeda operation dirty bomb infosec target juliett class prism activate!

I have had the displeasure of being on the wrong end. Amazon seller and vendor support is obtuse and designed to confuse. It is adversarial. It is about three steps away from being on the same level as a dealing with a federal criminal charge.

I had to write a plan, to avoid my 'misconduct', which was repeatedly rejected by 'assigned team', a group of unnamed individuals that was only allowed to communicate to me in automated templates.

I would only wish this experience on my worst enemies.

There's some assumption here that there's a reason for the account blocking, and if only a human could be reached, it could be explained in detail. But that may not be true. ML systems usually don't explain themselves by default; it's possible there's simply no further information available for a support agent to give.

This is a fun talk about post-hoc explanations from ML systems. It's possible, but it takes real work: https://www.thestrangeloop.com/2017/just-so-stories-for-ai-e...

Unfortunately it's all too common with Amazon to not give a great reason as to why where and how you violated their terms/expectations (IMO 3rd party sellers have to go down a support rabbit hole in the hopes of finding the one support person that can "guess" what's causing the suspension). Can you imagine how painful it would be to that third party Amazon seller whose whole income is from their (recently) suspended account?

Am relieved to hear that Amazon was not the author's only source of income...note the similarities to uber/lyft drivers being suspended.

This is as designed. All these companies ban or block users without explanation because their hand is forced. If they reveal the detailed reasons, or even make it possible to discover why, they become wide open to gaming. It’s a losing battle to begin with and keeping every advantage (secrecy) is critical.

Unfortunately.

It falls into the “this is why we can’t have nice things” category.

This is the correct answer.

Most of those "automated" replies were copy-pasted from humans. This book inadvertently brushes up against something they want prohibited on their platform, but if they explain why it will give you a vital clue as to how to game their system. They can't get into detail, but want to stop you. It looks like the author is writing stories that could be mistaken for Steven Universe fan-fiction, so it may be a lot more straightforward than this writeup lets on: The writing may come off as being related to a franchise they're not a part of.

Why doesn't Amazon at least include a hint as to what's wrong.

Automating this doesn't seem problematic per se, but omitting a key piece of info like that borderline malicious. It's a bit like an error message that just says "Error" - short of an absolutely green programmer I'd expect better.

That would open them up to people who can try to figure out what the triggers are detecting unwanted activity and getting around it, and possibly opening them up to discrimination claims.

Say what you like about Amazon, but you must admit that they are a very shrewd company. Provide fantastic customer service to the consumers, and almost none at all to the sellers. Brilliant.

Although they do sometimes fall down for consumers, I have a 2nd gen kindle bought from the US but running on the UK store which no longer talks to the store (I can download content but not browse).

I ended up in a maze of the US folks saying it's a UK store problem and the UK folks saying it's a US supplied kindle so talk to the US. After a few rounds of "reset back to factory" support calls I gave up.

I suspect the problem is the clock which is out by an hour so SSL / authentication isn't working. I held out until we left summer time back to GMT but even then it was still broken (I also tried from the Caribbean in case it was a European time zone issue). The GSM modem diagnostic page looked correct.

In the end I gave up and my partner got me a Kindle Oasis from Amazon UK for Christmas.

Did you maybe miss out on installing the firmware update for the store prior to the deadline? https://www.amazon.com/gp/help/customer/display.html?nodeId=...

Iirc they were disabling some old SSL stuff (either sha1 or weak ciphers) that the old firmwares relied on. If that's your problem you just need to manually update and it should work again.

I had the exact same problem with a Japan-bought Kindle Oasis used in the US. My Oasis got stolen right before I left for the airport, wanted one while on my trip so ordered one in Japan. I got the exact same run around trying to get special offers removed: US support says to ask JP support because it was bought there, then JP support says to ask US support because it's registered there. There was an additional problem that 3G only works in Japan for those Kindles.

I ended up sending an email to jeff@ asking for an exchange of the Kindle to a US one. Ended up getting a one-time credit for the cost of a new Kindle with 3G.

Kindle money comes out of Amazon's pocket. Any replacement or repair request approvals might negatively affect the reps appraisal.

If someone steals your account they are unable to look back more than two email addresses (even if the multiple changes occurred in under a 24 hour period). If you can't guess the e-mail the attacker changed it to you are SOL. Luckily Amazon decided to auto-fill with the new address so I could use that information, but otherwise they couldn't help me.

Great read, I loved the letter vs. spirit workaround she employed in the end.

Amazon is growing in a bunch of directions very quickly, so issues like this are bound to crop up.

I think the solution to this one lies in a business strategy (scale support staff at the same rate you are recruiting customers & sellers) or automation rule (i.e. flag and send to support staff if more than X replies have been sent && account remains locked).

It will be interesting to see if they can solve these issues in a timely and cost-effective manner or if certain arenas (i.e. eBook publishing) will be neglected in favor of more lucrative & automatable ones.

Amazon once froze my seller account for something stupid and it was impossible to contact a human or get it resolved automatically.

They got humans involved and fixed it pretty quickly once I explained why I was canceling my AWS account.

Welcome to Amazon! Look at what is going on with 3rd party sellers there; reliable, highly-rated sellers get their accounts indefinitely suspended for completely bogus reasons with no recourse. I guess they don't need them anymore so they don't care about their shoddy ML and its false positives...

On a related note to Amazon taking automation too far, I've noted that they seem to have some serious problems with categorizing products. For weeks and weeks now, in my recommendations are "Men's watches" [1] which are almost always comprised mostly of SSDs and flash drives. The flash drive in question's product page lists it as the "#1 best seller in Men's Watches" and my guess is that SSDs are close enough to a flash drive to also get pulled into the mix. It's frustrating to see large companies with offices full of developers making such public mistakes and then pushing broken technology further.

1: https://i.imgur.com/Sp7Yd5s.png

Similar to my experience with them: I had two accounts, one US and one German. After I moved back to Germany I asked them to remove my US account, as I didn't have a US debit card anymore. First they were able to do that, after several complaints and waiting. But then they also deleted my other account, probably by accident. A so-called double close, which is fatal on glibc, and in my case. Thankfully I've changed my mind about their services and am quite happy without any account.

I've been locked out of my Amazon account for a year now, for reasons I don't know. Amazon reps told me I'd hear back, but there's been nothing for months now. Fortunately, the credit card they have on file expired, and the ebooks I can no longer read are not so important to me. So, good riddance, I say.

Again, this is the trend I posted about https://news.ycombinator.com/item?id=17014500

What happens if Amazon decides to disable your account? Well, folks. This is no longer a "what if", but a "now what?!"

Sure, you can sue. But as many people here are pro-capitalist, they will assert a company's decision to fire users with absolutely no redress of the data that was decided upon. Personhood rights, and all that.

(No, I don't believe that companies should get personhood rights. And yes, I do believe that we people have the right to see what data they use to determine how they work with us, or not. The GDPR in Europe is a good start.)

Where does it leave everyone? Well, for these people, certainly not using Amazon AWS now. And I'm guessing that they can't even extract their data either.

Aren't automated system "Great" ?

Hmm I wrote you an email. I am a dev only though. Let's see what we can do ;).

Twitter froze my account today and I have no idea why either. Hate blockchain all you want, but I'm in pure love of it. I/We sorely need transparency and decentralization.

How would blockchain help?

"Twitter froze my account!" "Yep. I see here twitter froze your account at this time, in this date, in this transaction chain. Proof of work is right here."

don’t knee jerk on the throwaway blockchain comment. he clearly is using it as an example of decentralized systems. a quite poor example for his point, but still.

Why poor if I may? Sounds actually absolutely excellent imho.

That's how: https://www.reddit.com/r/ethereum/comments/87k9iz/presenting...

How does that help? There are thousands of well-meant projects from overenthusiastic but inexperienced developers. The "interesting project!" graveyard is overflowing. Nobody knows this project or cares apart from some enthusiasts, and that won't change. Never even mind the discussion about the purpose of the blockchain here. Even if there was a purpose, I'd say tweets are the last thing that needs to be archived all over the place.

The unique service of Twitter is that it is famous, not that it is able to distribute tiny snippets of text. Just like Medium. By using the brand name service you get an increased chance to actually be noticed and read in exchange.

So like Facebook. So let's not change anything then?

Alright let's do this: you don't change anything and stay on Facebook/Twitter indefinitely.

You will change you mind.

while i think her response was clever isn't it basically lying. like if you knowingly phrase something in a way to give someone a different impression from what you actually will later claim to mean its just a really clever lie.

Which part of her response gives an impression different from what she actually meant? I assume she read the rules and plans to comply with them. She checked that none of her books violate the rules, so she doesn't have to delete any of them to live up to the statement.

You are correct. I re-read her response and it looks completely reasonable to me on second reading.

Why not send the "I promise to follow the rules" email - since OP clearly desires to do so - and then follow-up afterwards instead of repeatedly ignoring the instructions given?

As they explained in the article, they cannot agree to stop violating the terms if they have no understanding of how they're violating the terms.

They did review them, and could not locate the violation. So they contacted Amazon for confirmation and wound up in this bot loop.

The email doesn't ask them to agree to stop.

It asks them to confirm they've read and agree to the rules, which they do.

Reply to the bot, then escalate. If the bot required wording like "I'm very sorry for violating the rules", that'd be a potentially different story.

Even without an explicit admission of guilt, you're putting yourself in a compromising position. Amazon made an accusation and then asked for redress, so if you give them that and were then involved in a lawsuit you'd have to explain why your actions (giving Amazon what they demanded to forgive you for rule-breaking) are inconsistent with your claims (you didn't break any rule).

So, add "Having reviewed them, I do not appear to have violated any of these terms and believe this to be a false-positive" to the wording they request, and send a follow-up to customer service.

Present in court, if necessary.

That is exactly what she did.

It asks them to comply with the rules and remove any existing publication that is offending the rules.

Given that they are blocked, because one of their existing publications allegedly violates the ToS, and given that they don't know in what way it's impossible to comply without removing all existing publications, which presumably they are unwilling to do.

> It asks them to comply with the rules and remove any existing publication that is offending the rules.

Sure, and given that they haven't violated the rules, they can honestly say they're complying with them, and have removed any violating publications (of which there were zero).

Send a follow-up to a human afterwards, rather than trying to debate a bot.

> Sure, and given that they haven't violated the rules, they can honestly say they're complying with them

Given that they don't know which rule they supposedly have violated the most they can honestly say is that they believe they are complying with them. It's not as if it's impossible that they have violated some rule in the eyes of Amazon, no matter how they read the ToS, and no matter how carefully they checked their own publications.

> Send a follow-up to a human afterwards, rather than trying to debate a bot.

This assumes that they know a human to contact, but strangely enough it also assumes that they aren't able to contact that human without first interacting with the bot.

There's kind of an implication that your are admitting guilt. That's probably the main reason she didn't do it right away.

Also, if she ever slips up on any rule in the future, Amazon can double-hammer her with violating their terms and breaking her own word.

Finally, by making it a link, Amazon could potentially change the contents and make it appear that she had agreed to something she didn't.

[Click here] to confirm you have stopped beating your wife.

Because he probably doesn't want to be seen to be admitting to a wrongdoing, if he does not believe he has committed any. And also because it is impossible to comply with a request to remove the incorrect material if it has not been explained what in particular is incorrect.

It's pretty-much a statement that you never want to make.

Please do yourself a favour and read "The Trial" by Franz Kafka.

Reply on news.ycombinator.com