The key point here that was hard to find is that they and others had a username containing Clyde set already, and now can't update anything else about their profile because their username has come to be considered invalid.
That's pretty poor judgement to not even communicate this to affected users
> That's pretty poor judgement to not even communicate this to affected users
Thanks for the clarification, but imagining it's pretty easy that this could just be a bug, e.g. they knew they wanted to prevent future usernames from having "clyde", and maybe even realized some existing clydes would be grandfathered in, but didn't realize it would prevent existing clydes from updating anything else on their profile.
yeah, I've had this happen a number of times because the service was treating the `+` as a space in either the signup or the update, and I've had some success forcing it by replacing `+` with `%2B`
This is a good case for why system bots/messages/etc. shouldn't be given regular names (and perhaps distinguished through other means too); if the error was instead "Username cannot contain 'discord'", it'd be more obvious.
At least in the naming, sure. Everything else, not so much. I had to switch my whole house over to Echos because you can't change the wake word on Google Home. Every time I said, say, "Hey Google, turn the bedroom lamp off", the Home in the bedroom would say ok and turn the lamp off, both phones in the room would activate and complain that they didn't have a device named lamp listed, and every other Home in the house would start talking about how it couldn't hear me clearly. Then you yell "Hey Google, shut the fuck up!" and the blasted things start whining about how they "have feelings too" and I shouldn't be so rude. Few technological problems have made my fingers ache more for a sledgehammer.
The developers of these things need to stop treating people swearing at them as a time to lecture them and instead realize that if someone is swearing at it then they’re probably not satisfied with how it’s functioning in the attempts before that.
Funny enough Google Assistant on Google Maps does this just fine. When it asks me if I wanted to navigate to the Pizza Hut that's five miles away or the one that's 38 miles away, it understands "the closest one, you stupid piece of crap" just fine.
I can see this being useful. There are multiple Krogers and Walmarts where I am. They vary greatly in quality and selection. Depending on where you are and what you want to get, you might not want the closest one.
Haha I wrote a HN comment years ago about the same issue. Everytime "do you want to go to the Wegman's 5 miles away or the identical Wegman's that's 20 miles away?"
This only happened to me once, but I was leaving BWI airport and asked Google to drive home. I really only needed a bit of help getting onto I-70 because some of the signs were confusing. Well it started navigating and after a few turns, it definitely seemed off. Turned out it was taking me to a Home Health place. I've always been super careful since then to make sure Home was Home.
Glad someone is, been trying to figure out if there was a command in Siri to communicate I'm dissatisfied with how its wasting my time with the past query attempts, no luck so far although I've basically given up using the thing at all at this point.
The amount of times where I would ask 3 times and have to do the task myself anyway were too often.
You can't change the wake word because a separate, dedicated chip is listening for it and is not connected to the rest of the device except to wake up the main processor.
Then they could at least provide more than one option. Though while far from perfect, one of the things I like about Alexa/Echo more than the ability to change it, is that with multiple devices you will generally only trigger one (the others within earshot will listen and then ignore it once one of them has picked it up; works well in my house at least apart from one place where echo/reflection sometimes causes one in another room to get preference, but even then only one device usually triggers). Sounds like that was the main problem for GP.
Most of the time I can just turn in the general direction of the device I want to "listen".
Yep. Echoes also don't make you pause after the wake word, so you can actually talk to it naturally. Also, when you tell them to shut up they shut up. The way Google Homes throw a tantrum when you're rude to them will make you want to reach through the cable line and strangle the arrogant son of a bitch who OK'd something that pointlessly infuriating.
Seriously, what the hell were they thinking? If I'm so frustrated at a device that I'll tell it to shut up, the last thing I want from it is more noise.
I've read a few times about the concerns people have with voice assistants, which all predominantly default to female voices, taking orders and abuse without question. I believe Cortana was also intended to "stand up for herself" in these situations, to avoid making it seem okay/normal to bark abusive orders at a "woman".
Great idea. When an aggressive man starts hurling verbal abuse at a robot with some female traits, let's antagonize him further with woke rhetoric from the safety of a remote datacenter.
With that sort of reinforcement, his next unfavorable encounter with an actual human female will surely end well.
Does nobody even try to de-escalate situations anymore? The most advanced AI in the world can't slink away quietly, wait until he calms down and try to correct these sorts of behaviors by promoting the idea that he'd slay it in bed, lose weight or grow a monster dong if he read this self-help book (now 20% off!) about treating women with respect? An incorporeal AI with goddamn control of a smart home can't walk through a hostage negotiation playbook to manipulate a violent occupant into submission? Clearly domestic violence is an impossible problem to solve.
It pains me to see people (usually women) encouraging vulnerable women to "stand up for [themselves]" in situations even armed male cops won't approach without backup from other armed men. It's sociopathic-- on par with teenagers encouraging other kids to commit suicide.
IIRC it always worked without the pause. Just the beep was picked up by the microphone and made the recognition for that word or two much worse. It would clearly react to things said before the beep.
Google devices function the same way when configured properly. The mention of multiple devices responding, and some not knowing about the light, makes me think GP has something set up incorrectly.
Dedicated wakeup chips can be designed to have a different matrix shoved in for the word they're detecting. Google presumably did not do this, but they could have.
Heat is dissipated when electricity goes unused too (usually by dumping it in some load resistor) so it’s possible the extra electricity is just generating heat, waiting to go to the processor. I’d lol if they have 5v coming in, and they’re stepping it down to 3.3v for the wake word processor (hence heat) while the main processor runs at 5v.
>Heat is dissipated when electricity goes unused too (usually by dumping it in some load resistor)
This isn't true. Your light switches do not heat up to lightbulb temperature when the light is off. Heat is only dissipated when electricity is "used", more or less by definition. You seem to be describing a linear voltage regulator, which does have a load resistor and is wasteful if the step-down distance is large, but no modern electronic circuit would use such a thing in a power supply - it would use a switched-mode buck regulator instead.
Switch mode power supplies don't waste that degree of energy when doing nothing. That's multiple watts, probably at least 5 judging by the temperature, just being pissed away at all times.
I guess that's why laptops batteries last exactly the same whether the computer is idling or not. Also, the reason they get very hot when you're not doing anything CPU intensive - it's all that unused electricity going to the dump load resistor!
You're installing a box that has a mic sending things you say intentionally and unintentionally (because it thinks you woke it up) to a megacorp. I think privacy is out the window by now.
That's interesting, I did some digging into Apple device communication and it turns out they have a protocol called CompanionLink that handles this exact circumstance - devices can communicate securely with each other and decide who is going to handle the Siri activation - HomePod's seem to get preference, then phones then laptops. I'm surprised Google hasn't done something similar.
I have no idea how things work under the hood, but this doesn't match my experience as a user. If I'm in the bedroom and ask to turn the lamp off, and phones are around, the phone screens sometimes lit up as they are listening to the command, but then only the bedroom Google Home actually executes it, while the phones turn their screen back off without saying anything.
Unfortunately I'm with the other guy. We have a Nest Mini in every room, and a Hub Max (only paid for 2, another 2 were free with Spotify and the Hub Max was an Instagram giveaway win).
We have to whisper to the Hub Max in the kitchen, from 3 inches away, and the Minis in the dining room and living room often still hear. If I don't whisper, even the ones upstairs hear. Sometimes I get 2 timers for my cooking, sometimes I get one on the wrong speaker. Then I try to cancel the timer, and the Hub Max hears instead and tells me I have no timers set.
The Hub Max asks on the screen if the wrong device responded. It clearly ignores your response, as the wrong device also responds the next time. I have pressed that button easily over 50x before I gave up and stopped.
Something clearly changed ~6mo-1yr ago because they never used to be this bad. It used to be that they would all listen, then only the closest one (that heard the command the loudest) would respond. Problem is, when I'm cooking the Hub Max is 1ft away and the rest are through (often multiple) walls metres away from me.
If this means you set each echo with a different wake word, you probably don’t need to do that. As long as each one knows about the other (same account) they choose a single device to respond based on the one which heard you best.
Google should have done that too, but it seems you might not have had them all part of the same home (or something else was broken/buggy).
Well, the verbification of the word google introduces its own difficulties there.
The ideal would be for the voice assistant to have some completely unique word, ideally with sound combinations that are very rare in your native language. And if for some reason, that still doesn't work for you, have a way to change it. Or maybe just not have a default, and require users to pick their own keyword.
My roommate had an Alexa and we changed the wake word to computer... Didn't last long. Apparently the word computer comes up on casual conversation much more than you probably think.
In English, there’s only one word in the entire language with a ts sound (pizza) that’s not at the end. They probably could have come up with a name utilizing that fact, making wake words super “easy” to recognize and hard to accidentally say.
Looks like I can search for " T S " to find the "ts sound that's not at the end"
% grep " T S " cmudict-0.7b | wc -l
991
Some of the non-proper nouns include ACCOUNTANCY, ANTSY, and ARTSY. I'll assume this is too close to the end to count, so require two sounds after the " T S ":
% grep -E " T S [A-Z0-9]+ [A-Z0-9]+" cmudict-0.7b | wc -l
853
(It doesn't have "spritzer" in the list, and uses the spelling "matzoh" instead of "matzo". I didn't look for a more complete list of word pronunciations.)
Finally, two words with two occurrences of the "ts" sound, neither at the end:
% grep -E " T S .*T S " cmudict-0.7b
ITSY-BITSY IH2 T S IY0 B IH1 T S IY0
TSETSE T S IY1 T S IY0
You're right. Though https://en.wikipedia.org/wiki/Boatswain says "formerly and dialectally also /ˈboʊtsweɪn/ BOHT-swayn", the dictionaries I checked do not list that alternative.
> Phonetic spelling bo'sun/bosun is attested from 1840. Fowler [1926] writes, "The nautical pronunciation (bō'sn) has become so general that to avoid it is more affected than to use it."
"Zeitgeist" seems like a bug, I've never heard anyone pronounce the leading 'z' as 'ˈts' in English, and doing so as a non-German might even be seen as 'over-pronuciation'.
Tzatziki, Tsetse and Tzar are not English words; OP said there was just one English word (which he identified as "pizza", which isn't an English word).
And "tsetse" has been used in English for nearly 100 years longer than pizza - the latter only really introduced because of Allied soldiers sent to in Italy.
For me, I apparently say "Hey, go [get, find, etc]" a lot and my phone wakes up unnecessarily. It seems to also happen with "Let's go". It's pretty annoying.
My personal conspiracy theory is that "Isis" is a result of the US "convincing" media organizations to use that name for the Islamic State after realising that intercepting communications using the term "is" was an impossible task. Even further, they then tried to impose "Daesh" but it didn't catch up.
Didn't they get death threats and insults on Facebook? Not only from people that would do that kind of stuff online, but also the kind of person that thinks that a terrorist organization has an official Facebook page.
They had to rename their page to "ISIS the band" or something like that.
Strong disagree. The whole "Karen" meme reeks of old-fashioned misogyny wrapped up in 2020s language. "A woman, complaining? How dare she!". I recognise the meme is about entitled women being rude, but that behaviour is hardly isolated to women.
Where's the simple line that gets crossed? If I ask for a refund but am refused, am I being "entitled" when I push it? There are some obvious examples of both ends of the spectrum, but a lot of less obvious ones in the middle.
I worry these memes encourage timid people not to even ask for what they are legitimately entitled to, for fear of ending up on some Reddit channel.
Yeah, it's all subjective and contextual. I don't know if I or anyone else can give an easy checklist to define what is or isn't "Karen" or entitled behavior, but having done my time in various positions in the service industry, it's certainly something I can know when I see it.
If I had to try I'd say it's when you have a problem and you try to resolve it in a way that gets you the most personal benefit and hormone rush instead of just getting the problem solved in an efficient manner, that's Karen behavior. Like if you order a sandwich with no tomatoes and it comes with tomatoes on it; if you call over the waiter, describe the problem in a normal tone of voice without assigning blame, and ask for a replacement, that's fine. If you use a tone of voice loud enough that other tables can hear, blame the server or cook or even take the tomatoes personally as if "ignoring" your request was a personal attack, launch into a diatribe bout how you could have bene killed if you were allergic to tomatoes, demand not only a replacement sandwich but a free side or even your whole order for free, etc, that's Karen behavior; those sorts of actions are getting you free stuff and serotonin bumps but they're causing undue stress on the staff and other people around you, who are human and have their own limitations in terms of stress and mental health, what they can do in terms of the law or dealing with their managers, etc - and they're certainly not solving the original problem (you want a sandwich without tomatoes) any faster.
As Scott says in Weak Men Are Superweapons [1], a meme aimed against a specific, easily disliked subgroup is unavoidably an attack on the entire group by association.
It is impossible to say "Karen, who is a woman, bad" without implying "women Karens, women bad." Or if there is a way to disentangle them, it certainly won't fit in a meme.
> a meme aimed against a specific, easily disliked subgroup is unavoidably an attack on the entire group by association.
So that comment, by attacking a specific, easily disliked subgroup of memes is undeniably an attack on all of the memes by association. Or am I applying this wrong?
I think so. If the argument is that people will make associations that the original person did not intend, then it seems quite possible that pointing out bad usages of memes will result in people likely memes less overall. This is but one factor and could be cancelled out by many other factors causing them to like memes more. I'm guessing this applies to humans in general and is caused not just by memes but by any transfer of information that biases schemas we use to judge situations. But this also applies to any terms of negative connotations where there is some subgroup associated with a larger group (even if the association is wrong, it just needs to be perceived). So even terms like 'racist' or 'incel' end up with collateral damage.
All right, let me explicitly substitute for pronouns: your comment was aimed against specific, easily disliked subgroup, namely, "Karen" memes, and so it is unavoidably an attack on the entire group, namely, memes, by association.
It is impossible to say "a "Karen" meme, which is a meme, bad" without implying "memes "Karen" memes, memes bad." Or if there is a way to disentangle them, it certainly won't fit in a comment.
It's amazing how many arguments ad argumentum are not only self-applicable but also self-defying.
Sure, if you arbitrarily decide every comment is a meme.
It's amazing how everything is applicable to everything if you freely redefine the terms used.
edit: Sorry, that was a bit snarky. I think there's reasons why this applies to quick, punchy memes but not to longform comments: memes are antithetical to nuance.
Perhaps that hints at the problem of intentionally developing a meme stereotype, and then using it as a stick to beat people with. The bar for having that stereotype becomes ever lower, and thus the set ever bigger. People then self-censor perfectly acceptable behaviour to avoid being part of the set.
Karen is a _white_ woman who uses their entitlement to attack people in services.
I like how you just white washed the problem of racism with misoginy here.
That whole group of Karens is probably a product of white male misoginy towards white women so these women found of way of coming back at it by directing their anger towards even less priveleged groups.
Mental health is more important than telling yourself that adapting makes you weak. And please stop with this bullshit along the lines of "just ignore it", that's not how bullying works.
Heavily bullied as a kid. Never conformed. Maybe not the right solution for everyone but I learned those were not the types of people whose approval I needed in life.
I hope not, my daughters name just happens to have syllables similar but not the same as “siri” in her name and calling her name sets it off ALL THE TIME.
Your data is about "percentage of female names in whole population".
I'm talking about "most popular name to assign to newborns".
Sorry for misunderstanding.
Alexa is not such an uncommon name. They should team up and sue, being designated as the obedient servant for the benefit of someone else's brand is a non-negligible annoyance. Ignore people who say 'tis' not that serious,' they're the sort of people who never care about anything until it happens to them.
I also feel bad for that Google kid I knew in high school. People were always asking him where to find stuff. He thought that changing his name from Alta Vista would stem the tide.
Oh at first with my cafein lacking early morning brain, I mistakenly parsed it as "Developers should not be able to call their bots regular names". But it's nothing to the bot's name but how it's registered on the system right? So the suggestion is, the username should be discernible from platform users usernames.
> This is a good case for why system bots/messages/etc. shouldn't be given regular names
I disagree. The username can be clyde_discord/clyde_bot and the display name could just be "Clyde". Then the username Clyde wouldn't be taken, and the users would be able to still message the bot and not be confused.
Though I personally really dislike simple human names for bots because it sounds cheesy and dumb.
It's not about the username being taken, it's all about what people appear as in the client; bots and scammers might impersonate Clyde to steal credentials, social engineer, etc, so you cannot be "Clyde_help" nor "Clyde_Official" nor any variation that would make people think the bot/user is speaking on behalf of Discord.
Why not just put a special logo or symbol or something on that profile to mark it as special/official, or have some sort of "user type" displayed that says "human" or "official bot" or "third party bot"? I feel like making it look like a regular username makes it _more_ likely people will be scammed by "other" regular users rather than less likely.
Discord already does this - they just don't trust average users to be aware enough to check. (Based on my experience moderating some discords, they aren't wrong.)
Which is funny when you take into account that people are still using special characters to get a Clyde name that mostly resembles it. I question the effectiveness.
There's not exactly a list of canonical human names per se, but "bots can't use human names" would still be a very sensible rule for us to have. Put the cost and the burden on companies making bots, not on individual people who got unlucky.
Also names are part of human culture, and Alexa for example has been in use for a very long time. There are valid reasons for names to die off in a single generation (eg adolf) but this is not a reason that our culture gains anything from imo.
There are certainly far more Clydes than there are Discords (has anyone even been legally named Discord before?)
I'm not happy about it, but those are the rules.
..."then change them", as someone might say. But that article seems to be mostly about the edge-cases. I don't think someone named Clyde is an edge-case, nor would someone named Bonnie, for that matter.
If your plan in choosing a name for your bot is to choose one that won't collide with a human, you are going to run into the fact that humans have a lot of very different names. Hence the link to the 'falsehoods' piece.
The US Census of 1950, for example, has a lot of Descords, Descards, Dicords, Dioscords, and... in PA, a Rosa B Discord, in NYC, a Mr Discord.
In general, all I'm saying is, it is far better if your intention is to allow people to choose their name, to not put any arbitrary restriction in place with the justification that 'it probably won't affect anyone'. If there's one thing people really care about it's their name. And no matter how unlikely you think it is that someone will run into your restriction... you're almost certainly wrong.
> has anyone even been legally named Discord before?
There was a Roman goddess named Discordia (where "discord" came from), and in the 90s Hercules and Xena TV series the Greek goddess used that Roman name instead of Eris, but simplified to just Discord.
Those two series were pretty popular, so I can see it being more likely recently than in that 1950s census sibling comment found. On the flipside, that character is probably not something parents would want for their daughter.
"Clyde is the name used for a fake bot in the client. It responds to specific Built-In slash commands (/timeout, /ban, /kick and /nick), and sends private messages with information or errors". https://discord.fandom.com/wiki/Clyde
It's not. The reason for this is not technical (unless you think they literally filled their user database with every possible username permutation with the word 'clyde' in it). It's to make sure scammers can't use names with these words in it, because it might make more people consider those accounts legitimate.
Yeah, sure. The very first comment in the Reddit thread says "I use Clydē and it works," and the third one is "CIyde. Use capital i instead of L," so nice try, but no.
All usernames on Discord have a four-digit number after them called a discriminator, separated from the rest of the name by a # symbol. In the Reddit post in question, that number is #6381.
The problem isn't that the (fake) bot is in the same namespace as users; if there were any problem like that at all, it would be a shortage of discriminators.
However, that's not what this message is saying. It's saying that the name cannot contain "Clyde". In other words, you wouldn't be able to use any name that had the word "Clyde" in it, taken or not.
The fact that they haven't attempted to block usage of "CIyde" and similar does suggest they might not have thought it through, but it's verifiably not a namespace issue.
I think you're talking past each other. Having to try to block names containing or appearing similar to "Clyde" is a result of having the bot in a public namespace.
Just as a random, poorly thought-out example, internal stuff could be namespaced with a different prefix like "&Clyde" instead of "@Clyde". As long as users know that official stuff is addressed with &, it matters less what people do with @-namespaced stuff because there's an easy way to tell whether it's actually official or not.
By putting official stuff in the same namespace as user-generated stuff, it's much harder to tell whether something is an official tool or not. @CIyde could be a real internal account.
It's probably a mix of both.
Namespacing often seems to be an afterthought in many systems, and it can be hard to refactor in once an architecture has solidified.
It's not always the wrong choice to mingle identifiers of different types, but I think often people err on the side of convenience (/ laziness) instead of thinking through all of the potential issues.
The discord username namespace is weird. You get to set a display name which doesn't have to be unique so you can have the name "Bob" and that's what shows up when you message people, but then you also have a unique version on your profile that looks something like "Bob#3922" I'm not sure what happens if 10,000 people pick the same name but as far as I know, there is no limitation to how many people can have the same username.
So it seems almost impossible that this is some technical limitation but rather Discord cracking down on a common deception tactic.
It's really embarrassing if does actually allow visually indistinguishable variations like 'CIyde' or 'Clydе'. Impacting real users, while having zero effect on malicious actors
It feels like a UX faux pas to pick common names for stuff like this.
I’ve got to imagine that some people named Alexa find it unfortunate and invasive that suddenly their name is used everywhere. At least “Siri” and “Hey Google” are less popular names for people.
On Swissair website there was a time where I couldn't purchase tickets because my name contains the name of a musical instrument. So out there there are people that did worse than that in my personal experience :(
Wait, what? I'm trying to figure out why that would exist but the only thing I can think of is that people were buying a ticket for their instruments rather then checking them and the airline wanted to try and stop the practice.
I was shocked as well, and couldn't come up with an explanation either. My name is Marcello and the problem was the `cello` substring (an Italian name, but being Italian an official language in Switzerland they should have though about that). They removed it luckily!
That is the instrument that I had in mind when I was thinking of the ones that musicians buy tickets for, but I couldn't think of a name that had a collision with it.
I had heard that some airlines may have tried to discourage the practice but this is the first I heard about such measures.
It's a common security practice, to prevent scammers from using official-looking usernames, or even tricking services that verify domain ownership by sending email to postmaster@ or admin@.
I discovered it innocently. I wanted to create a gmail address named support{myfirm}.gmail.com. Of course I instantly understood what you point out but then I had to discover all the other edge cases.
So you think it's bad? How about this: I just wanted to create an account there finally and it was suspended without me even entering it once, I also had to pass about 5 captchas to be able to sign-up, no VPNs, clear home IP
Some shit-show is going on there right now, dunno
Contact forms for US Senators and Representatives often have surprising issues like this.
Reserved words in SQL sometimes get blocked as well as things like too many singles quotes or an odd number of single quotes. I've seen the name "Walter" blocked on many forms over the years.
The worst part is that the sender just gets sent to a security or generic page with no idea why, often losing the message they created.
This feels very adjacent to the Scunthorpe problem, except that the blocking is intentional here:
> The Scunthorpe problem is unintentional blocking [...] because [...] text contains a string (or substring) of letters that appear to have an [...] unacceptable meaning. Names, abbreviations, and technical terms are most often cited as being affected by the issue.
Actually, in this case it's intentional that "Clyde" is blocked. That's the name of the "bot" that error messages on Discord use to make them friendlier.
It's obviously not a good solution, but it's not an example of the Scunthorpe problem.
That's pretty poor judgement to not even communicate this to affected users