[UA-EAI] some notes Re: EAI Reading Call

Mark Svancarek marksv at microsoft.com
Thu Apr 12 17:42:29 UTC 2018


Agree that explaining motivations can clarify a text in some cases.

My comments:
--  Punycode is typeable, but also phishy and error-prone.  There is no reason to believe that they can easily be used by anyone in any language. 

-- Anecdote: I previously owned Product Keys for Microsoft (those 25-alphanumeric stings that one would manually enter when installing or activating software).  We found that increasing the number of characters from 15 --> 20 --> 25 generated a massive increase in support calls from people who typed them incorrectly.  The error rate, already high for English-speakers, was *much* higher outside of the "ASCII-zone".  Removing visually-similar (allowing "B" but not "D", disallowing both "I" and "1", for example) reduced the error rate, but not entirely. 

-- Many of us believe that most EAI usage will be between people and entities which most commonly use the same scripts.  Inability to read or type or pronounce the characters and labels is reduced in such cases.

-- A goal of UASG is to encourage large corps like Lufthansa that they should allocate resources to dealing with these issues.  Large corps understand that adding such support implies investments elsewhere (updating a website involves not just web developers, but content creators, support staff, and IT staffers and LOB application developers). Your example of phone support having to understand how to process macrons is something that would plausibly need to be updated as well.  

-- These days the scripts that the phone support folks work from are increasingly augmented with AI, MT and bots, so it's not unreasonable to think this is manageable.

-----Original Message-----
From: UA-EAI <ua-eai-bounces at icann.org> On Behalf Of Paul Borokhov
Sent: Thursday, April 12, 2018 10:06
To: Arnt Gulbrandsen <arnt at gulbrandsen.priv.no>
Cc: Dina Solveig Jalkanen <icann at thomascovenant.org>; Ram Mohan <d8f546ce-cfbb-4a97-91f8-e7905fd8b341 at pexch112.icann.org>; ua-eai at icann.org
Subject: Re: [UA-EAI] some notes Re: EAI Reading Call

As Jiankang said I don’t want to re-argue this point, since I’m not convincing anyone, but the document should explain the motivations for those being introduced to the topic for the first time. See below.

> 12 апр. 2018 г., в 12:45 ДП, Arnt Gulbrandsen <arnt at gulbrandsen.priv.no> написал(а):
> 
> Paul Borokhov writes:
>> Isn’t this just an implementation detail? Why should (average) users be exposed to it?
> 
> It's an entrenched implementation detail.
> 
> It's entrenched in both software and average users. The secretary at my children's school is a nice example. She has an Excel sheet with all the parents' email addresses. She expects that she can update that sheet with addresses she sees in parents' mail and also that the addresses are readable and understandable over the phone.
> 
> Lufthansa's customer service expects the same thing. Both that the addresses they see in my email are readable to me over the phone, and that an address I read to them over the phone is usable to send me mail.
> 
> And so on and so forth.
> 
> Jiankang wrote that some readers may prefer punycode. But that doesn't matter to senders, because some other readers reject punycode. Some customer service representative would see Don's new email address (from the attached screenshot) and say "Mr, Hollander, I'm afraid there appears to be an arror with your email address. Is tee oh aye at ex enn hyphen hyphen enn que pee you kay aye pee you kay aye dot enn zed correct? Do you perhaps have another address we can use?»

What’s the alternative? «Mr. Hollander, did you say your address contains the letter z with a line over it? I’m sorry, I’m afraid I don’t even know how to type that on my keyboard». How would I explain to a Lufthansa representative that my email address is борохов@почта.рус?

Punycode is typeable and readable on any keyboard in any language on a machine that’s connected to the internet today (it has to be). Anything else isn’t.

> Authors of software must assume that sending a punycoded address risks being rejected by some recipients.
> 
> Arnt
> <don-hollander-datamail.png>

_______________________________________________
UA-EAI mailing list
UA-EAI at icann.org
https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fmm.icann.org%2Fmailman%2Flistinfo%2Fua-eai&data=02%7C01%7Cmarksv%40microsoft.com%7C3eeec806bebe4abd9d7808d5a097afa4%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C1%7C636591495710146715&sdata=JiImhgNFwgsLg7w9yKdOdFn1YpK%2Bdgrd2iSk%2BxZgxQw%3D&reserved=0


More information about the UA-EAI mailing list