I asked Patrik Fältström from SSAC (since he mentioned the presentation before the Tech Day) and he says that it is a work-in-progress of SSAC that might result in a
report in 6-12 months.
He also says that in that work the use that looser definition of homoglyphs. He refers to
https://en.wikipedia.org/wiki/Homoglyph
The SSAC presentation shows that we should not take the risk of confusion too lightly.
I think this also illustrates that if we use the term "homoglyph" in the strict sense we should also state that in our report.
Mats
From: "Tan Tanaka, Dennis" <dtantanaka@verisign.com>
Date: Thursday, 25 October 2018 at 15:54
To: Mats Dufberg <mats.dufberg@iis.se>, ICANN Latin GP <Latingp@icann.org>
Subject: Re: [Latingp] Variants
Hi Mats,
I did attend the ccNSO session on IDN Homographs. While the presentation was given by SSAC members, the information or analysis were not produced by SSAC. They used an “off-the-shelf” third-party’s data, so
I would take the information with a bit grain of salt. I did not find a SSAC report.
I found their definitions of homoglyph and homograph a bit loose. The dictionary defines homograph as “one of two or more words spelled alike but different in meaning or derivation or pronunciation (such as
bow of a ship, a bow and arrow)”.
As far as “very similar”, I read this as “nearly identical due to font design”. The example they use (a and ã) would not fall in the very similar, but distinguishable category, in my opinion.
-Dennis
From:
Mats Dufberg <mats.dufberg@iis.se>
Date: Thursday, October 25, 2018 at 2:34 AM
To: Dennis Tan Tanaka <dtantanaka@verisign.com>, "Latingp@icann.org" <Latingp@icann.org>
Subject: [EXTERNAL] Re: [Latingp] Variants
Dennis,
Thank you for adding those bullets. At the first glance I read them as ANDs but of course they should be read as ORs.
Speaking of homoglyphs, SSAC had a presentation at the ccNSO Tech Day Monday. I did not attend since it collided with some IDN meeting, but I have looked
at the presentation.
In the presentation they have one slide on homoglyphs and homographs. From the slide 10:
◉
Homoglyph One of two or more glyphs with shapes that appear identical or very similar
a ã
https://static.ptbl.co/static/attachments/191691/1540208800.pdf?1540208800
Please note "appear identical", not "are identical". And also "very similar". The message here is quite different from the first message from IP.
Please view the presentation and especially slide 17.
I have not looked for an equivalent SSAC report.
Mats
From:
"Tan Tanaka, Dennis" <dtantanaka@verisign.com>
Date: Wednesday, 24 October 2018 at 19:56
To: Mats Dufberg <mats.dufberg@iis.se>, ICANN Latin GP <Latingp@icann.org>
Subject: Re: [Latingp] Variants
Mats, thanks for the summary. I believe it’s an accurate account of our action items.
On #4 there’s some additional nuance. In regards, similarity vs variant, they pointed out to limit considerations of appearance to cases that:
-
are unambiguous
-
have overriding security concerns
-
exhibit true exchangeability (homoglyph)
They also mentioned that when looking at similarity cases we should look at those through the lenses of a “careful and observant user”.
-Dennis
From:
Latingp <latingp-bounces@icann.org> on behalf of Mats Dufberg <mats.dufberg@iis.se>
Date: Wednesday, October 24, 2018 at 12:48 PM
To: ICANN Latin GP <Latingp@icann.org>
Subject: [EXTERNAL] [Latingp] Variants
Friends,
During the ICANN63 we had one specific session between Latin GP and IP, but there have more IDN sessions where comments relevant to our work has come up.
1. IP wants us to create cross-script variants with all code points in different scripts that has identical or visually similar to LATIN SMALL LETTER O.
IP gave the following list in their presentation today:
Code Glyph Name
006F o LATIN SMALL LETTER O
03BF ο GREEK SMALL LETTER OMICRON
043E о CYRILLIC SMALLER LETTER O
0585 օ ARMENIAN
SMALL LETTER OH
0B20 ଠ ORIYA
LETTER TTHA
0D20 ഠ MALAYALAM
LETTER TTHA
101D
ဝ MYANMAR LETTER WA
IP that a GP has the responsibility to look at other GP's proposals, both completed and work-in-progress, for obvious cases of visual similarity.
2. Besides the "ring" above, they want us to look for the straight line, i.e. l like.
3. IP suggests that we include cases of visually similarity that we have decided to reject in an appendix of two reasons:
Firstly to show the reader that we have considered the case, but decided not to included it as variant. In that way we might get less questions during public
comment if the we have looked at certain pair.
Secondly to give data for others, such as for a 2nd level IDN table creation.
I know that we have decided to do it, but I just want to emphasize that it came up.
4. I feel that there is more understanding on creating variants on visual similarity rather than strict homoglyphs. Other GP propose such in-script variants.
Mats
---
Mats Dufberg
DNS Specialist, IIS
Mobile: +46 73 065 3899