Hacker News Reader

Comment by latexr | Hacker News Reader

> rafaelconde: Are you a person or a bot?

> Kyla: you are currently talking with a live support

That did not answer the question!

Seems reasonable for certain non-native English speakers. "You are talking with a live support [person/agent]", but support is a noun meaning "support person". Chinese developers who have introduced themselves to me usually say, "Wu Chengming, develop" instead of "developerER". (The name is made up) Also, the strange construction using the indefinite article is a feature of some non-native English.

This doesn't mean it's not a bot of course (TechSupport Syndey seems likely to claim to be a live person), although I think a bot would tend to have more commonly used grammar.

In general, people who learn a language later in life are far more prone to errors where the language has two or more words that all map to one word in their native language.

Chinese does not modify words--thus things like develop vs developer are going to be a problem. Likewise, they only have gender on family and on written pronouns--thus the problem with Chinese speakers being bad at gender.

Sometimes people are given a stupid script and have to say things like that. It's fun to try and get them to break character, harder on chat than on the phone. Or it could be a chatbot

If they're a human who has to follow such a tight script that they can't even tell you they're a human, then they're effectively a robot anyway.

This is so true.

Sometimes even people without scripts are merely trained on a handful of expressions, and they just rotate through them.

I can't believe their words are written live, by humans. They must use decision support systems that provide them with text bricks, with "write custom answer" being one of the options.

My personal favourite is when you go to open a support chat and it makes you enter your issue into a text box. Then the chat starts, you repeat your issue to save time, and then regardless of the issue sitting on their screen twice, they ask what your issue is. I assume it’s either automatic or they’re penalized for not clicking the button that “says” it.

I always assumed so as well, but that kind of suggests they either intentionally misspelled "restriction" as "restruiction" to make it seem like you're actually talking to a human that is writing their responses, or they didn't, and nobody has ever noticed, or the tickets don't ever reach anyone who has the ability to fix the error.

I don't know. The replies are strange and sound like they're not written by a native speaker. If they had pre-selectable reply-blocks, wouldn't those be written by professionals?

I know a household brand who put up an ad on a freelancer site for $2 an hour for dev work. So there's that. On top of that, according to "The Design of Everyday Things" >>no one cares<< how things fail (e.g. copy of an error text or support message) so lesser functions, especially cost centers like support are almost always going to be off-shored.

That makes sense (well, not the $2/hr dev work), but even if you outsource the actual support, wouldn't you build the system they use?

Amazon's support that only barely understands me and sounds strange (feels like they do german <=> english translation both ways and theirs is worse than google translate) leaves a bad impression on me, so at least I care, but I'll happily admit that I'm both no one and also very much not the average customer.

Still, it just feels wrong. Granted, I've never run a support team, but I expect most of it to be repeated things. Someone's package hasn't arrived. It has arrived but was damaged. It wasn't damaged, but it contained the wrong product. Or the wrong amount of products. Or the product wasn't new. Or the product is broken, or is the wrong color etc etc. And there's not really a lot of variety in it. You find out the issue, you offer one or two solutions ("money back or resend item"), you tell the customer to have a nice day. Having pre-made blocks of well-written text to choose from (you can even have the agent choose in english and the customer gets the localized version) seems like a trivial thing, but it'll transform the customer experience from "okay, I guess Amazon only employs imbeciles" to "wow, what a nice person, and they were typing very quickly".

Large corporations are confusing to me. I'd expect someone at Amazon going "yeah, let's do that, it'll save millions and improve customer satisfaction, it's no-brainer". Yet they don't. And so I wonder: do they know something I don't ("we've tested this, people are more cooperative if the support agent is obviously non-native"), or are they just incompetent?

I visited a contact center for a major-ish online retailer once for work, and watched their processes. They had people dedicated to chat, who would operate 5 or more chats simultaneously, and in general were typing rather than copy/paste iirc. I'm sure it happens though, these guys I visited were pretty low tech.

It's often easy to see the copy paste variety happening though, e.g. in email discussions where you get sent some wall of text that's barely applicable to your situation. I've had this happen a lot with amazon and uber.

I don't mind the pre-programmed walls of text, but please don't make them too big. I got a wall of text answer out of Garmin--which promptly scrolled much of it off the screen. Unfortunately, what was left (in the window they constrained the size of!) was a coherent whole that was completely irrelevant. The relevant part had scrolled off fast enough I didn't realize there had been more.

(And I never did get anything useful out of their support, but they ended up fixing the bug.)

Yeah, it's a script. It sucks when the script is bad though. Like one time on the phone with some company I have a card with, I was asked for my "first and last name" when an agent picked up. I answered correctly and the agent shuddered like I said something wrong and was asked for my "first and last name" again. After a while I finally answered incorrectly and gave my name as it appears on my card (which includes a middle initial and suffix in addition to my "first and last name"). I explained to her the difference between what she asked for and what she really wanted, but she didn't seem to be happy with the feedback :(

The scripts are incredibly annoying. I would rather them just short-cut to "I cannot do anything useful for you at this point, and I cannot tell you anything that would help. There is no value in continuing this conversation."

This isn't an acceptable excuse.

Is the world we want to live in really the one where pursuing support from a company requires the end user to perform a Turing test on their platform's support interface?

yes it did, they implied they're a bot... but bots have personalities, they are persons, just not biologically embedded

Reply on news.ycombinator.com