I've worked in high volume customer services. It's not easy for the agents doing the work or the companies operating them.

I've worked in multiple call centers dealing with calls and tickets. I've worked in companies in which I dealt with mobile phone bill queries and hardware issues right up to Cisco networking problems and then eventually enterprise grade Linux clusters at Rackspace (UK). I've spanned a pretty large spectrum of industries from a customer services perspective. I've even done customer services in the food industry.

It's HARD. Not just for you as an operator, but as a company trying to offer the service to begin with. of all of the companies I worked for, O2 (a mobile phone provider and network operator) had tens of millions of customers. We had several call centers that spanned close to ten thousand agents doing everything from calls to tickets, emails to Tweets. The call queue for that company never dropped below a constant 200-300 on hold and emails took days to even get to.

Google has approx' one BILLION customers.

What is your suggestion here?

> they are mostly trying to automate the whole process as much as possible which then causes errors like this

Damn straight automation, to a larger degree, is the answer. People don't scale at all. They can do one job at a time, for a limited amount of time, and are subject to all kinds of problems.

> Yeah I know it's terribly expensive to keep a horde of content reviewers

It cost O2 about $5/6 to answer a single call. That's the cost of the agent's time and JUST answering the call, not the cost of listening, understanding, resolving and ending the call. Humans are expensive.

I'm not saying don't use people. I'm saying automation at Google's scale is one of the few solutions available to them when it comes to dealing with the simply insane amounts of complaints they'll receive just via their YouTube platform alone.

EDIT: Apologies for being so crude in my original wording. I'm updating the comment to reflect a more professional, civil tone.

Google's customer service is horrible even for organizations that pay them +$10k/month or make them +$10k/month for ad sales, and that is a far smaller number.

You would think they could at least scale customer support properly for that much smaller number, but they don't because culturally they do not want to. Even people who have used google cloud services say the support is bad compared to AWS. Google stands out as a corp that is really bad at customer service.

While apple, amazon, AWS & many other companies provide customer support that mostly works for customers who provide maybe $40/month max in revenue. They don't even let you self pay for your own support at google in most cases.

I bet youtubers who live on their work on youtube would gladly pay $100 to deal with one off incidents like these that might happen once a year if it meant they had proper thought applied to their issues for example.

Is this true? As a paying customer for gsuite, the 4 times I've contacted customer service I got someone immediately and they solved my problem immediately. I'm not a big customer either I have one account. I pay them $72 a year.

I have a grandfathered free GSuite plan on a domain I hardly use.

The few issues I have had have been promptly solved by a real person.

It's very much opposite when it's come to AdSense, but again they don't really care when some webmaster lose his source of income since there will be other webmasters to replace him.

Fundamental difference: person who buys adds --> Google gets money person who uploads videos --> Google gives money

It's pretty clear to me what the antithesis is here. Sorry to over-simplify, but yes, imho this matters.

i have dealt with google europe, i even visited the offices in dublin but i do not have a confidence that they know the answer, most of them are there to try to convince you to try this new thing and incrase you budget.

   > i have dealt with google europe, i even visited the offices in dublin

So obviously my impression might be limited to quite superficial questions or problems. It's very much possible that if you work with something actually complex their support isn't anywhere as useful as it's for entry-level customers.

"Absolute certainty" aside, OP basically just said they saw the mountain from the base (only has the view of the volume of tickets pending in their queue) and most of it is in a cloud, they can't even see the top, it looked big, hard to drill, impossible to shave down to ground level. The "absolute certainty" confirms the point of view.

Senior management see it from above (see the big picture) and see total volume of work, a mathematical division based on resources, the value of each work area, which are core or not, where to focus more resources, and which parts to cut down because they set the end goal.

The customer sees the paid promise of a flat terrain. No mountain, no resources.

Now I don't expect the support engineer to know or care about more than what comes in their queue. That's their only job. But I do know management almost always sees support areas, especially for the "low value/impact" segments, as the best candidates for some savings when you have shareholders to please. Successful management will strike a balance between saving more money from cutting costs than they are losing by losing customers and image. Anything else is considered shortsightedness: they either leave money on the table, or they choke the business for short term gain.

In this particular case YouTube could solve the issues but they have deemed this not worth it. The amount needed to fix this is higher than the return so it can't be justified. Fixing one offs and high visibility cases is cheaper than eliminating the issue entirely.

Paying or not, people will complain.

I'm not saying my knowledge applies to all customer support cases, but what happened here would never have happened with our system.

Just very simple tricks go a loooong way dealing with malicious people:

how old is the account of the "reporter" ?

is the "reporters" id verified?

how long has the account we're taking action against been our customer?

whitelists/blacklists -- temporary or permanent

are all complaints against a customer of the same type grouped to the same ticket type/ticket?

Just based on those simple facts, we would mmediately lock functionality/content for unverified, new customers we received complaints about, while delaying, notifying and giving verified, aged customers the chance to respond/fix issues.

I'm not saying this is easy, for the system I worked, our agents designing the rules had about 80 individual ticket types(in a hierarchy, with rule re-use for ticket category), but they used their brains and did not take any "lock X functionality of the customer" lightly.

My point was we had automated rules in place that worked really well and there was nothing really clever about those rules.

It was a game of cat and mouse between our rule designers and adversaries who always tried to probe our thresholds and evade us while we tried to find new useful facts about customers, new probes, new actions, react proactively, but not screwing our legitimate customers was always our number 1 concern.

How hard can it be to augment crowd sourcing and "AI" with some actual humans? If the "AI" is so crappy that it can't figure out an established, popular account is worthy of human support, what problems can it actually solve?

I also wonder why we can't just pay $X to get a human, but I know the answer. Tech companies would just take the money as extra profit and still provide the shittiest quality support possible.

But maybe, there isn't, currently. Maybe there simply isn't a solution at the scale of companies like Google.

If you double your customer count, and double your revenue, then you can double your customer support resources. O(n) does not become impractical just because the number gets really big.

If they can't deal with something, the fundamental design is wrong, and would fail even if they downsized a thousandfold.

If you have 1 person doing support, they can deal with the weird cases themselves. If you have 2 people, they need to share information between themselves. If you have 10 people, they might need to set up some sort of escalation system. If you have 100 people, you'll probably set up some sort of routing system (multiple languages/specialization). If you have 500 people, you probably need some sort of internal abuse / anti-bribery effort to catch bad apples. If you have 1000 people you've now got a separate team working on setting up policies. If you have 5000 people you've got a training and evaluation system set up for your customer service staff. If you have 10000, you've got customer support staff in multiple countries around the globe and need to deal with conflicting and ever-changing laws in the countries you do business in.

All of this while marginal revenue per customer declines as your scale goes up. So yes, in addition to it being costlier per customer to provide the same level of support as your customer base grows, you have a smaller budget per customer to do it in.

I don't know about that. Just a few bad actors going after a small site can be an utterly ridiculous percentage when compared to the abuse a site like youtube gets.

> If you have one person [...] If you have 10000

All of that sounds sub-linear to me. If 10000 support staff require 1000 direct managers and 200 people working on policies and training and 100 people working on development and 100 upper managers, that's still O(n).

> All of this while marginal revenue per customer declines as your scale goes up.

If some of your customers are higher revenue, and some are lower revenue, I'm not sure if that's really a scaling problem. But even with that consideration, it's not an excuse for being unable to give proper support to paying google customers, or youtube accounts with 100k subscribers.

All of the things I mentioned are things that you can get away without but eventually scale forces you to tackle. Each of those things brings with it additional cost without increasing the throughput.

This shouldn't be too novel an idea: consider a normal supply curve: cost per unit tends to go up with volume.

That doesn't mean it costs more than O(n).

Let's say you start off with 100 workers providing 100 units of support, and 20 workers of overhead.

Let's say doubling support means you need double the existing workers, plus double the existing overhead, plus 20 more workers to handle new kinds of overhead.

So 120 x 2 + 20 = 260 workers for 200 units of support.

Then 260 x 2 + 20 = 540 workers for 400 units of support.

The percentage of overhead grows every single time we double the number of workers.

And yet if you do the math, you see that the overhead never goes over 40%. And it was already 20% when we started, so that extra amount isn't hard to afford at all.

Despite growing every stage, it's still O(n). The dominating factor is linear.

Obviously those numbers are just an example. But the point is that you have to look at how much the costs increase. Even if it's theoretically faster than linear, if the overhead only grows a moderate amount as you grow from a thousand support workers to a billion support workers then it's not an excuse. It's within a small fudge factor of linear for all real-world numbers.

> This shouldn't be too novel an idea: consider a normal supply curve: cost per unit tends to go up with volume.

Only in the short term. In the long term, production ramps up until the price is pretty close, and on top of that economies of scale drop the price by a significant amount.

Even if you could double your customer support resources, it means a very different structure and needing to ensure you both hire in the same quality of employees up and down the escalation chain as well as adapt your organization to handle the vastly larger communication and problem resolution processes.

Noe of that is trivial and it's quite easy to take a great org and make it terrible by just adding a bunch of people.

Sorry, but you can't skip support because it's not cost effective, just like you can't skip financial regulations or data regs. You don't get to use the excuse of "Sorry we are too busy making billions, maybe we wouldn't have to cut corners if we weren't so successful."

The problem is that good customer support costs more than $1. It costs more than what they make from the average user, so the average user can't have it. That would still be true if they had 100 times less users. It's a result of the service being free, not the service being large.

There aren't a billion content creators on Youtube, and even fewer who would be good targets for blackmail like this.

The channel in question has 100K+ subscribers and 17M+ views. That's probably a sizable amount of profit for Google from ads. It it really so unfathomable that google could tune their ban bot from "ban this video/channel/account immediately" to "signal a human to have a look and decide on what to do, maybe after talking to the channel creator"? That's just ridiculous to me.

Your treating the instant case as the common one, but it's not. And you don't know that until after you've done an investigation, which means you don't know it when deciding whether to allocate resources to an investigation.

It's trivial to game subscribers and views. If having more gets you something then all the actual garbage channels will have plenty and you're back to square one.

Google gets something like two million takedown notices every day.

If google cannot even make a bot to detect fake views and subscriptions, then why the hell do advertisers give them money at all if only those bots watch the ads anyway?

And why would they think that a ban bot - which has far reaching consequences - is ok, if they cannot even prevent view gaming?

It's not trivial to game subscribers and views. Such gaming is a problem, and they are pretty surely not really close to perfect detection rates, but they aren't that bad either. And everytime I dare use a public VPN google/YT puts big fat captchas in front of me.

>Google gets something like two million takedown notices every day.

Not impressed. Not every "content claim" leads to a takedown, or account strike. And not every content claim is the same. If they cannot distinguish between one coming from e.g. Sony vs one coming from e.g. <random script kiddy> then what the hell are they doing? There is such a thing as reputation.

Because the ad slots are sold at auction. If 20% of the views are bots then the price should go down and advertisers get 20% more ad views (and the same number of real ones) for the same dollar. Assuming the advertisers are rational and informed; if not then the explanation for why they do it is that they're irrational or uninformed.

> And why would they think that a ban bot - which has far reaching consequences - is ok, if they cannot even prevent view gaming?

One thing doesn't appear to have anything to do with the other.

They have a ban bot because people keep trying to sue them or threaten them with adverse legislation if they don't take stuff down fast enough.

> And everytime I dare use a public VPN google/YT puts big fat captchas in front of me

Because you're doing the opposite of what scammers do. You're using the one IP address with a terrible reputation. They use a million IP addresses with good reputations, e.g. from a botnet full of machines that were innocent users last month.

> If they cannot distinguish between one coming from e.g. Sony vs one coming from e.g. <random script kiddy> then what the hell are they doing? There is such a thing as reputation.

You would think so, except that a ton of the erroneous takedowns come from the biggest senders, because they use garbage automated methods to send them.

Meanwhile if they give privileged access to major companies then everybody else would be complaining that they're discriminating against small content creators by not honoring their takedowns like they do for Sony.

[0] https://www.bangkokpost.com/world/1909764/alphabet-google-pr...

And it would open up a new trolling vector. Now the troll files complaint after complaint and forces you to pay $100 over and over again.

What could work is to do loser pays. If someone wants to object to your video, they have to post $100, and so do you if you want to get it reinstated. Then whoever wins the dispute gets their money back. And extra points for having a builtin troll deterrent.

Once upon a time i've been a happy customer of a regional ISP called Hansenet, started by the local public electric utility. It was (at least for me) lightyears ahead of the crappy experience (speed/latency/price/(competence of)service(if needed)) the former state monopolist had. Everything was golden, also excellent peering. They grew slowly, cooperated with some other regional upstarts, stayed golden, speed even got faster without raising of monthly fees, furthermore no long contracts necessary, could be canceled within a month.

Then Telecom Italia came and bought them, now it was Alice, as advertised by some mouthbreathing twen all over the public space, larger than life.

Quality of service went down, very unsexy of you, Alice! But still was better than the former state monopolist, and still got faster without raising prices or contract conditions.

They even retained a local service center where you could walk into and talk to people who knew their stuff and could change settings in realtime! Imagine that! Not some outsourced call-center agent mouthpiecing some useless uml-diagram.

Anyways, too good to be true? Right! Telefonica came to rescue (of BlackRock), to milk the cash cow as long as possible, and it was noticable.

Fortunately another neighboring public utility, pioniering fiber roll out since the nineties expanded and became a choice for me then. All is golden again! Yay!

Why am i ranting? Because everybody i know is mostly unhappy with O2 in any form, except for the price.

O2 can't do!

I wish i could add some internal, even worse insights from another ISP growing larger at the time, but am not allowed to. Also they managed, got better, and meanwhile can do again.

Lesson? There is no lesson here, with the exception of economies of scale are bad for humans above certain amounts of 'scale'.

If those are the stats and you're still not actually solving many customers' problems; if tons of people are being hung out to dry with no recourse, then you're doing a bad job. If you don't have enough people to do the job, then hire more people. If you can't afford to hire more people, then don't do the job, or limit the number of customers you deal with.

There is also the scenario in which there is no solution and you need to balance the cost of investigating to the point where you're 100% certain there isn't a solution and efficiently deciding there isn't one.

One thing I have learnt working at these companies is that some customers think the world revolves around then and that technology should work flawlessly all the time, when it doesn't.

That's why every mom and pop store outranks every Fortune 500 on customer service expect Amazon. Most still haven't learned to put customers central to their business but treat it as an after thought.

I like to use Telecommunication companies as an example because I worked in one heavily and dealt with these exact scenarios.

If a customer calls the service desk because their internet is broken due to vandalism at their exchange. I as a customer service agent cannot drive to the exchange and simply fix the problem myself. It requires appropriate prioritisation of all of the other problems within the network that need fixing that may also be affecting other customers, as well as requiring outside resources to rebuild that location. To add that the customer might also have a mobile SIM as a backup within the modem but is not able to achieve the speeds nor latency needs commensurate with their fixed service. And that's if you're lucky to be the wholesaler if you aren't, then your organisation literally cannot fix the problem legally because you don't have the authority to repair infrastructure that you don't own outright.

I appreciate that this a specific example, but one I have dealt with mutltiple times. Once a business grows to a greater size, it is inherently more difficult to fix problems like a mom and pop store whether or not you care the same about your customers or not. It may appear as though they don't care, but there are simply different realities at different scales.

My point being that if you assume as the beginning that they purposely distribute available work and responsibilities in such a way that to execute is difficult by design then that's not what we're talking about here. What we're talking about is where you are operating at such a scale that it is very difficult to provide great customer service even if you do care.

I worked alliance support for a major database vendor in a former life. My experience is that enterprise customers are usually realists. They run a very complicated environment and are aware that things, especially software, is not perfect. They expect good service (and pay through the nose for it) and can be demanding. But ultimately they're realists.

The ones that think the world revolves around them and their issue must be fixed RIGHT NOW! are usually the ones that bought some ODBC driver for 40 bucks 12 years ago, which don't really have any support entitlement to begin with.

I know, blasphemy to the VC crowd.

Customer service at scale, for many products, turns the product from profitable into a severe loss. It will go out of business no ifs ands or buts.

YouTube clearly cannot survive if its customer service costs exceed its advertising revenue.

It has absolutely nothing to do with speed of growth, and I don't know why you think it does.

Are you saying free user-generated video hosting sites shouldn't exist period? That YouTube simply shouldn't exist? Because that seems to be what you're saying.

Either have proper human or machine evaluation and very fast response to complaints about fuckups, or,

let it be an open platform, and only respond to legal requests to remove things, the way the Internet used to work,

...because they clearly cannot handle the alternative without fucking people over repeatedly.

If that is the case then Google is helpless to stop these attacks from growing out of control. How long before the long tail of YouTube is clipped off?

That's what algorithmic detection (together with user flagging) and removal at scale is for. That prevents them from growing out of control.

The combo of algorithmic detection at scale, combined with small amounts of manual intervention and customer service where most needed, is how YouTube is able to operate.

This article is evidence that the process is breaking down. It's an old maxim in computer security that attacks become more sophisticated over time as knowledge of the system grows and spreads among attackers. That is what we are seeing here. YouTube has built a system which is ultimately unsustainable at this scale. Over time these attacks will continue to escalate until only the very large channels (those they have the resources to support in person) and the very small channels (those beneath the notice of attackers) are left.

No, the article is a single instance. If it becomes common, YouTube will create systems to prevent this type of attack.

> YouTube has built a system which is ultimately unsustainable at this scale.

There's just no evidence of that -- that's pure conjecture. And given how Google has seemed to manage abuse and spam at a commercially successful level in plenty of other products (not perfect, but maintaining the platform), I would bet on YouTube continuing to succeed here.

Consider that you're out of touch.

Right now Google makes the second most profit per employee of any large enterprise on earth: $269,000 in profit, annually, per employee.

I know of other- successful!- companies that make less money per employee, because they hire people to do customer support.

To start your argument, I think you have to explain why Google can not hire more CSRs, even though they are the #2 most profitable enterprise on Earth.

To begin your "back of the envelope math test", you should ask:

If Google were to hire CSRs until its profit went to $0 per employee per year, how much of an effect would the additional CSRs have on the experience of Google's customers? How many CSRs would that be? How many customers would each one be handling per day?

Saying that a company which is profitable because a small number of employees serve a large number of customers can obviously afford to hire a large number of employees isn't a math test, it's wishful thinking. Do some math.

It would go up, probably quite a bit, and the profit would re-emerge but at a lower level per employee.

The real problem is that their scale (in terms of reach, money, customers, political leverage - you name it) is so large that a company could compete or out-compete on customer service and still wouldn't win a significant chunk of their business. That's why they don't need to hire more CSRs.

You need to answer that one before concluding that customers' experience would improve.

So no, I don't need to answer that beyond any number greater than there are now, which I already did.

Or let's make it 3-fold and claim the other hundred grand is for training and office space (yeah, can't get a CSR to work remotely and definitely can't earn less than minimum wage nor those jobs be abroad because I'm building an unrealistic quibble).

It's like people arguing that you can't know the sky is still blue on Earth because you haven't been outside yet today. Please. Worse than that, it again obscures the real questions - what level of CS does Google need to provide to stop customers going to a competitor, which is the level they have now, and hence, since they're too powerful to be competed with what should be done about it? What can be done about it?

Apparently temporary works and contractors make up around 54% of Google's workforce. https://www.vox.com/2019/4/4/18293900/google-contractors-ben...

The rest of this comment is going to be somewhat critical of Google criticism and seemingly apologistic. Please assume positive intent.

I agree that Google certainly could do more to help people on its platform. But this will always be the case until every single customer service issue is solved to the user's satisfaction. The important question that I never see anybody asking or answering is "How much is reasonably enough, and how do we know (subject X) has done enough?"

It all boils down to what we believe the moral obligation of people (and companies, by extension) ought to be. Any extra service that Google provides beyond those required for profit maximization (e.g. customer retention, maintaining reputation, being only slightly better than the second-best option) is purely charity. Then, how do we evaluate when a certain amount of charity is sufficient, let alone necessary? Public sentiment is fickle, and therefore not a great measure. I bet even if Google increased its CSR by 10x, we'd still see communities eagerly latch onto any criticism of the tech giant and shout "you have to do more!" without quantifying what the final demand actually is!

I take the moral position that we as a society certainly can and should do much more to take care of each other. Telling each other to be better is absolutely valuable and I will never argue against that. But I believe we ought to do a much better job of suggesting desirable and realistic end states, instead of perpetuating insubstantial discourse amounting to one of either "They need to do better!" or "Come on, they're doing the best they can."

Admittedly this places me in a non-committal position. I don't have any great suggestions for what the end states of YouTube, Twitter, et al should look like from a content moderation perspective. But I would love to hear input from actual experts in who can offer some reasonable ideas about the most cost-beneficial solutions for society as a whole.

No, it isn't.

But the damage it causes to innocent parties through its choice of business models is.

It is easy to forget that not so long ago, the Internet giants of today were little startups too, and not so long before that, they didn't exist at all. They have reached the dominant positions they are now in through a combination of factors. Several of those factors involve absolving themselves of responsibility for things going wrong, even things that might have been (or might still be) considered illegal, because preventing them is difficult or practically impossible with a purely technological solution.

Well, guess what? There is no law that says they get a right to exist and be astonishingly profitable by relying on technology at the expense of anyone else who gets hurt by their action or inaction. There is no law that requires the existence of huge video redistribution services that allow anyone to upload content, including malicious content that may cause serious harm to others, with impunity. Laws that do protect these organisations, such as the safe harbour provisions under the DMCA and its counterparts around the world, were invented after the fact and it's far from clear that they strike a healthy balance (particularly when, as I can personally testify, the likes of YouTube do not always meet their obligations even under the very generous terms of those laws).

If these companies can't act as responsible corporate citizens using their current business models, maybe they shouldn't use those business models. It is disturbing that this simple idea now seems almost heretical, as if we are somehow beholden to these Internet giants and the world would fail without them, and so it's somehow OK that they also magnify harmful effects from spreading dangerous misinformation to invasion of privacy but as long as it's someone else uploading the content the hosting service is deemed innocent of all wrong-doing.

Duty is not charity. It would be perfectly reasonable for society to determine that YouTube incurs a duty to its content creation partners, and at no point should that be considered requesting a favour.

I agree that Google has a moral duty to take care of the people who choose to put their trust in their platforms.

See https://en.wikipedia.org/wiki/Business_judgment_rule

But, point taken.

That’s not the logical conclusion, that’s the inverse.

If P (large profit per employee) then Q (should be able to spend more on customer support)...

- The converse is If Q then P (clearly not applicable in this case)

- The contrapositive is If not Q then not P (in this case, you would hope it’s true but obviously not!)

- The inverse is If not P then Not Q, which would roughly translate into if profit per employee is low then you would not expect a company to pay more for customer support.

Perhaps you could argue that the logical conclusion is that any profit just means they aren’t spending enough on their support. Funny thing is economists would argue profit demonstrates they are spending enough on keeping customers happy.

It certainly appears that they can afford to do better (although arguably they don’t break out financial results just for YouTube do they?) and we certainly want them to better.

I think the fundamental problem is that for every million dollars they pay humans to “do better” it’s like trying to move the beach with a pair of tweezers — you don’t even make a dent. No amount of money will solve the problem by trying to pay humans. They can only succeed by making the algorithm better, and that’s not a problem which scales with headcount (mythical man month)

> That’s not the logical conclusion, that’s the inverse.

Your logic is misapplied. If Google's profit is an indicator of its customer service obligations, it is logically necessary that less-profitable companies have lower such obligations than more-profitable companies do. If they had higher obligations, Google's profitable status would indicate that it didn't have much in the way of customer service obligations. If they had equal obligations, then profit would not be an indicator.

I simply wanted to point out that a statement like "Companies that are wildly profitable should not shirk customer service responsibility" is not logically equivalent to saying, "Companies that are not wildly profitable should shirk customer service responsibilities."

To argue that because you disagree with the inverse of a statement, therefore the statement itself is not valid, is faulty. The inverse of a statement need not be true for the statement itself to be true.

"No company should be let off the hook for having awful customer service" does not preclude judging one of the most profitable companies in the world more harshly for having awful customer service.

What I read from OP's statement is mainly this; we know in this case that it's not for lack of resources. So perhaps we can agree every company has the responsibility, but not all companies have the means. Is it not more reprehensible to fail to provide a service you are responsible for when you have the means to provide it than when you do not?

Or more to the point, it's certainly noteworthy that they could become one of the most profitable companies in the world despite having such widely recognized terrible customer support. One might say it's significant evidence that they are able to maintain their profitability through not entirely competitive means.

> "No company should be let off the hook for having awful customer service" does not preclude judging one of the most profitable companies in the world more harshly for having awful customer service.

What you're describing in that second pulled quote is a continuum of customer service responsibility versus profitability. According to you, the more profitable company has a greater responsibility than the less profitable company does.

E.g. "I couldn't hire more customer support agents because I'm broke," and "I am one of the most profitable companies in the world, but I still refuse to hire more customer support agents because I'm a monopoly and don't have viable competitors so it doesn't matter anyhow"...

No, this is incoherent. If the judgment for your failure to provide support is "that's understandable; we could hardly expect you to do any better" -- that is the same thing as you not having a responsibility to provide support.

I also probably should have more charitably interpreted GP as "companies in a strong position have a better ability to provide good customer service."

Which is fine, because Google's customer base is made up of people, so Google should be able to scale their support in a linear way.

on edit: I don't mean that not supporting their user base is the right thing to do, although an MBA might say it were.

That was easy, thanks for asking the question so that I can answer it.

Any more questions you think "we" should answer? I'll be glad to resolve them since you only seem to ask them.

That's an insane statement. YouTube should shutdown because they're successful?

What, do you think a monopoly is a bad thing? The market place decided YouTube was the better platform and continued to pile on (sure, they had market timing too.)

If you can suggest something better than YouTube do let me know.

When things go really wrong on our office365 platform or any part of Azure, Seattle will call us on the the hour with updates until it’s fixed. We have programs within Microsoft support where our developers and sysops can suggest changes, and if the suggestions gather enough popularity and fit within whatever criteria Microsoft have internally they’ll happen.

AWS didn’t always have the great support they do now, but as far as cloud operations (the nerdy bits) go, I think they’re actually better at it than even Microsoft. It build up gradually along with GDPR, but they were the first of the major three to offer things like support for AWS services that’s actually located within the European Union. I’m not updated on Microsoft, but AWS may still be the only company to do that.

Compare that to Google, where we get to talk with an automated system, the same as though we weren’t a 3,3 billion a year budget organisation. I’ve had better personal support for my free google account than I have for our company google cloud account.

Don’t get me wrong, we would probably have a lot of GDPR related reasons to prefer Azure or AWS even if google had better support, but the lack of support means we’ve never gotten beyond that step in our risk assessments.

I work pretty exclusively with AWS these days as a CloudOp. I remember building an environment in GCP and getting answers from support was terrible. I eventually left that job after 11 months.

The next job was pure AWS and sweet Jesus was the contrast strong. AWS not only had people who would phone you within 10-15 minutes, more often quicker than that, but they had a frickin' account manager who visited us every month and we only spent a thousands per month with them. Absolutely worlds apart.

Which means they also have one billion potential workers. They could tap into the power of crowd-intelligence. It's probably somewhat easy to forge some processes to oursource certain support-problems which need low human interaction.

Something they do with Recaptcha already and IIRC at Google Maps. They could sell Youtube Premium to some thoudsand people and let them instead check a handful videos per day for some cases of copyright infringment and such. Just let them be the final fonrtier of sanity in cases where the machine is weak.

They'd all have to be copyright lawyers, then?

If you aren't making enough money on YouTube to be able to afford paying someone to host your content and provide better, paid customer service, then you have to consider that the risk of being on a mostly automated, poorly supported platform is the only thing keeping you from losing money, or in other words, is the sole source of your income.

When you are as big as Google, very improbable failures happen all the time, pointing them out doesn't tell the story of how well they're doing unless your expectation is constant perfection. Go find rates, not anecdotes, then you might have something.

Also, if you're being blackmailed and harassed, it would seem to be a matter for the police and the courts, not Google, to fix.

Just a thought experiment, does Youtube need to be as big as it is? Can’t Youtube say “given the current accuracy/precision of ML based solutions and how much our manual operations can scale, we are going to rate limit the content that circulates to match our ability of giving the best judgment on what videos to take down and what videos to keep”. Physical stores have maximum occupancy limits, and with coronavirus some even shown proactive rate limiting to say “hey, we can’t ensure transmission control above a certain density”. Those are corporations too, those also have fiduciary duty.

I think “growth at all costs” is a major driver of these types of failures, especially when costs are externalized and distributed to the userbase; one annoying notification, one abuse feature change, one harmful policy decision at a time.

> Go find rates, not anecdotes, then you might have something

Youtube has near perfect knowledge of the warm data of their operations, and use it to optimize user frustration rather than minimize it, while we only get to know others are being abused when news make it to Hacker News. The inability to find the rates is inherent to the problem. Short of sneaking into Youtube data centers, there is only so much data you can find without Youtube’s say so. Which brings the question, for such a massive policy layer operation, why is this data not made transparent?

A few scattered moderation complaints compared to wholesale rate limiting of incoming content... I can only imagine the madness of content producers if YouTube tried that.

>Youtube has near perfect knowledge of the warm data of their operations, and use it to optimize user frustration rather than minimize it

YouTube content moderation isn't human spaceflight. Having a goal of never failing is, well, a very inexperienced way of thinking about failure. On the order of half a million hours of new video content a day goes on YouTube, or something like 6 million uploads.

Let's just pretend that the moderation decision is only made once on upload, you make 6 million decisions a day, how many ridiculous decisions are you going to make every day?

Do you think that number is ever going to be zero? The reality is that there is an exponential relationship.

Let's just say you can double your cost to cut failures in half. If it's not that relationship it'll be something like it.

You can just keep doing it. Doubling your cost over and over, and you're still going to fail all the time. When a "one in a million" chance means something happens several times a day, well, you have to have a different attitude about failure.

And when you're just about the only game in town, whatever you do to mitigate failure is going to attract people trying to screw with you to upset you and your users, to take advantage. Not only are you trying not to fail every day, you're trying to avoid people exploiting those attempts. It's just reality that you have a popular service, there is going to be some nonzero number of very upset users with just cause to be upset. That's just reality.

I'm sure there is already some per-user upload rate limits in place to prevent gross abuse. It is a matter of playing with these parameters. Rate limits don't have to be distributed mindlessly equally globally. Just like in Hacker News or in Twitter, certain proxy variables like the account age and karma scores are used to assess abuse potential and certain capabilities of the platform is unlocked to the user. Most probably something like this also exists. Again, it is a matter of playing with these parameters.

Additionally, "few scattered complaints" suffer from selection bias in telling what is actually going on. Not all policy (mis)applications transform into a complaint. If I had to guess, the ratio could be at least an order of magnitude but obviously we have no data on this.

> Do you think that number is ever going to be zero?

Zero complaints is a straw-man end-state. Of course there is going to be diminishing returns. You can aim to minimize user complaints at 80% to save from diminishing returns, or you can aim to maximize platform usage and ad revenue and take user frustration into account only to the extent it affects platform usage maximization. You can operate on either goals while paying attention to diminishing returns.

Very simplistically, there is a relationship such that "for every x dollars we don't spend on reducing abuse, we lose y dollars in user frustration". Who decides the satisficing ratio between x and y? This optimization is done asymmetrically by these types of services and in general there is no way for us to say "wait a second, turns out you were abusing your users/letting them get abused more than you should while making a smidge more profit out of it" with such information asymmetry.

Yet that's what the grandparent comment is suggesting:

> Can’t Youtube say “given the current accuracy/precision of ML based solutions and how much our manual operations can scale, we are going to rate limit the content that circulates to match our ability of giving the best judgment on what videos to take down and what videos to keep”.

YT already does this abuse-prevention, they just lean towards letting more content on rather than less. The grandparent suggests that they start limiting uploads in a way that more regular uploaders (eg. a few videos a day) would feel, which would warrant some explanation or statement from the company.

Come on. It is more than a lean. Growth is an explicit, fractal goal at every level.

> The grandparent suggests that they start limiting uploads in a way that more regular uploaders (eg. a few videos a day) would feel

I didn't give implementation details on the rate limit, because as I preambled, this was a thought experiment, not a design document. They whole point is to underline the fact that, there is a cost of making the review process better (whether automatic or manual) and to what extent corners should be cut versus this cost should be paid seems to be informed by the objective to grow at any cost.

The comment of yours I was responding to

>use it to optimize user frustration rather than minimize it

It is not at all a straw man. You are criticizing nonzero optimization and encouraging minimization. To minimize is to have a goal of zero. What other definitions do minimize and optimize have in the dichotomy you are trying to make here?

That's not what minimize means at all, hence my straw man argument. Minimize means reducing the amount as much as possible while still satisfying other parameters of your equation. Zero makes the whole operation nonviable as we agree due to diminishing returns, so it doesn't satisfy that definition.

The dichotomy of optimizing user frustration and minimizing user frustration is basically a question of what parameter gets what valence in the whole objective function. Maximizing profit at the expense of user frustration is optimizing user frustration, meaning having the perfect amount of user frustration in which their leaving the platform and the cost of reducing the frustration is on balance. Minimizing user frustration means doing this the other way around and having a product that, as Google claims, serves "the user first". If you understood that to be zero, I hope it is clear now. At any rate my point doesn't hinge on what the minimal point is. It is clear that any goal other than optimizing user frustration will be an improvement to current situation as far as what users get out of the transaction, and it is unlikely to be adopted.

> basically a question of what parameter gets what valence in the whole objective function

You're splitting extremely fine hairs there, your difference between minimize and optimize seems to be not much more than how much you like a term "satisfying other parameters of your equation".

>Minimize means reducing the amount as much as possible while still satisfying other parameters of your equation.

You have to stop after "as much as possible". Adding a "while still" makes it optimization, your value judgment doesn't get to determine what does and does not fit inside the "while still" of minimize.

Once somebody calls me out for a logical fallacy based on their own difficult to understand definitions of seemingly common words... if that can't be resolved there really isn't a point in continuing, it's not like we could really communicate anything much less make arguments if we can't agree what minimize means.

I'll try to simplify it. Imagine a function with dependent and independent variables. Choosing an independent variable so that the dependent variable is the minimum is minimizing. Imagine the plotting of the equation, we are still trying to be on the curve. That's what "as much as possible" means. You are saying minimizing has to be when that dependent variable is zero. That is only possible if the equation crosses the x axis. For our case, it doesn't because diminishing returns make that impossible. Hence my calling strawman.

> You're splitting extremely fine hairs there

The thing I've been trying to separate is not optimization and minimization because minimization is a particular type of optimization. I'm trying to distinguish what particular variable is given valence while solving the whole system of equations. Because in a case where two of those variables are correlated (user frustration & revenue), picking one over the other chooses different optima for both (optimizing for user frustration vs optimizing for revenue yields different optimum values for their correlated counter-parts; for revenue and user frustration respectively). I want to emphasize, these variables are not independent with respect to each other. So there is no simple "revenue is max and frustration is min" solution for which we could say "youtube should just optimize the whole thing". There is picking to be made and max revenue is picked over min frustration, always.

Picking up on me using words "optimize" vs "minimize" for two different variables appears to me as further strawmanning. I don't ascribe malintent, but what we have been discussing has been irrelevant to the conclusion of the original post from the beginning.

In other words, if you can't afford to buy cars and hire drivers and comply with all regulations then don't compete with Uber. The missing part is, Uber intentionally cuts these things out of its business model so you can't… I guess it would be, drive for Uber until you can afford a real taxi medallion.

Likewise, I think being a YouTuber you will NEVER be able to earn enough money that you can jump ship to a competing video service with better conditions, because no such service (?) exists at the scale and audience of YouTube, because YouTube runs at a substantial loss in order to completely own the market for non-trivial uses.

That's valuable. How valuable? Stock market valuable: as in, people with money THINK it is valuable, and I don't think the reality has been well tested yet. We're still in the world where dominating and monopolizing is seen as super valuable, because of some immense perceived payback that you apparently get to do.

That said there would be some privacy concerns if that data was just floating around out there.

Of course that might create another incentive for Google not to provide any customer service at all until people pay.

But YouTube is still a platform you must be on if you care about expanding your viewership.

Of course it's not easy. You have to figure out growth for yourself, and most indie businesses and projects fail. Then again, so do most YouTube channels. So it's up to you to choose the indie path or choose YouTube.

I don't remember any of the Internet idealism of the 90s mentioning the great progress of sharing information involving the creator of every piece of data available making a fraction of a cent every time it was consumed.

I don't mind people getting paid to do things of course, but I do mind a culture dominated by selling each other's attention to advertisers in exchange for creative output, or creative output solely for the dopamine shot from attention and likes.

I don't see that at all. Encouraging people to do things for their own merit and not just to make money – and we have to be realistic here, most people trying to make money on YouTube or anywhere by creating content won't ever see a single dollar from it, so it's more like trying to win a crappy lottery.

It's this constant sense of entitlement or talking like people are forced into making videos for a living which bothers me. It's a crappy job that is pretty easy to know how it works before you get into it. I keep seeing "no choice", well you can always do something else.

I guess I don't like the entitlement/blame cycle where people who help create a problem complain about the consequences.

Be the change we want to see in the world. Stop posting on YT/FB/IG/wherever, and create your own site, attract your own audience.

I'll come watch you/read it/listen to it. I'm sick of FB/YT/IG/wherever.

I have seen that, particularly in history channel circles were the content get demonetized for things like saying Hitler, many of them band together and publish to youtube last or not at all and promote these other channels in their videos. I don't know if any of them will be successful, but staying on youtube simply isn't an option for them either.

I think if/when youtube falters it will be death be a thousand cuts, it won't be a single movement with organization and it won't be to something else as centralized.

If one is not willing, capable, or even interested in trying to solve problems they help create, as an individual or a group, I have very little interest in sympathy for their difficulty.

Refusing your pet idea isn't a refusal to do anything at all.

At some point, people need to realize what constantly wanting things for free is costing them and the people they watch.

It's not free, we pay for the content by buying products, it's just very very indirect.

Which is the Star Trek model. It's the future of 'nobody has to work to eat, and so now people go insane trying to pursue popularity since survival's a solved problem'.

That makes it interesting in its own right, and we can already see both the possibilities and the 'race to the bottom' factor that ensues.

None of that changes the fact that Google is indeed "treating it like a fun ML problem that they want to solve with magical algorithmic pixie dust". That's basically their MO, and they put wayyyy too much stock in the results. Saying that Google doesn't do this is crazy. This is why the Podcast Addict thing happened; they were trying to crack down on the misinformation from "plandemic" weirdos and Podcast Addict and many other apps were hit with the spray of the shotgun blast. Is their original intention worthwhile? Yes. Do I think a human "reviewed" Podcast Addict before they took it out for linking to content about literally the world's most talked about story while at the same time Google Podcasts did the exact same thing? Absolutely not. If a human reviewed that case and decided that Podcast Addict (the #1 podcast app on Android) needed to be suspended, then something is broken. Maybe someone needs to be fired, but there's no someone in the chair. You can't fire the AIs at this point- You can only tweak and rework them and hope they take out less innocent bystanders.

All this said, is it hard? Yes. Is there too much content to really do this right? Yes I believe there is. Could Google do a better, more efficient and transparent job of cleaning up the mess when their scripts misfire? Oh absolutely yes.

We tried to convey the message [1] that due to corona virus fewer and fewer human reviewers will be available.

[1] https://www.businessinsider.com/youtube-coronavirus-content-...

Also you can’t expect people to put their heart and soul to this kind if grunt work - reviewing content and complaints all day long.

In the end it might be that humans just become robots on task like this.

And this, by far, is not the worst case of what's proliferating there.