String Similarity and the New gTLDs – Does ICANN Have Too Many Similar Extensions in the Running?


One of the most pressing concerns with nTLDs is whether or not the launch of so many in such a relatively short amount of time will complicate the functionality of the Internet for everyday users. Wrapped up in this worry, is the focus on string similarity, or how close a proposed nTLD can be in meaning, sound, or visual representation to another, or to an existing gTLD.

There are rules in the ICANN guidebook concerning how similar strings can be, and they seem to spell-out that strings should definitely not be close in visual representation, while giving more leeway to how close two strings can be in meaning and sound. As experts watch the process, and the upcoming Initial Evaluation results, which will indicate string similarity issues, some believe that there’s a lot of flexibility, while others believe there’s not.

ICANN defines the purpose of their String Similarity Panel as such:

“The String Similarity Panel will review the entire pool of applied-for strings to determine whether the strings proposed in any two or more applications are so similar that they would create a probability of user confusion if allowed to coexist in the DNS.” (4-3)

The guidebook further explains that the String Similarity Panel will look for the flowing issues:

  • Whether or not a proposed string is too visually similar to another string, either already in existence or proposed.
  • Whether or not a proposed string is too similar, by proxy. For example, if A string is too similar to B string, and B string is too similar to C string, then A string will need to be accessed as well, even if it’s not similar to C string on its own.
  • Whether or not the launch of a certain nTLD will cause user confusion.nTLDs domain names

In a letter written in November, but only recently posted to ICANN’s site last week, Jeffrey Smith, the CEO of Commercial Connect, LLC, let ICANN know that their guidelines are both hard to parse-out for interpretation and may mean that only 56 proposed nTLDs are actually unique, if the guidelines are taken strictly.

Smith, in conjunction with Commercial Connect, is applying for the rights to .shop, a follow-up to an application Commercial Connect submitted in 2000, when they were asked to submit again, as a part of the larger nTLD push. Smith states that his letter is written as a community member more than an applicant, and he points out that if all 1,917 applications are considered in accordance with the string similarity guidelines put forth by ICANN, then not just visually similar strings should be knocked-out of the running, but also similarly sounding and meaning strings, too:

“There has been an understanding that no new gTLD would be released that had similar meaning, sound, or appearance to any existing TLD. This has roots in protecting the end user from being confused about which TLD should be used and lends credibility to the intent of the process. The premise of keeping the internet a safe, secure and user friendly environment for all stakeholders supports this rule.”

To illustrate what Smith is talking about, here are just a few examples of current applied-for nTLDs that could be considered too similar to launch, depending on how strictly you interpret the guidelines:

  • .auto, .car, .autos, .cars
  • .loans, .loan; .market, .markets; .work, .works
  • .buy, .shop, .shopping
  • .lawyer, .esq, .law

There are many examples like these – where words are similar visually, many times just the difference between singular and plural form, or that words are too similar in meaning, such as with “buy” and “shop.” Smith argues that IDNs should also be considered in the same mix as other nTLDs – since both will be existing together as options to one another.

Yet, some argue that Smith’s representation is too oversimplified. If an applicant applies for .auto, and wishes to use it as a closed registry, in conjunction with their business, than a .car would suffice to establish an open-market nTLD in place of the word “auto.” Some argue that having “car” and “auto” in any capacity would be confusing; others argue that Internet users are flexible and capable of discerning the uses of both nTLDs.

With so many applications in the works, it’ll be interesting to see how ICANN broaches the subject, and whether or not we’ll see the number of possible nTLDs lower as launch dates get nearer.

Where do you stand on string contention? The more the merrier, or too many cooks in the kitchen? Use any outdated and corny colloquialism you want – we won’t judge you. We just want to get your take.

If you have your eye on the application process, check out our nTLD watcher. Plugin the nTLD you want to track, and we’ll send you updates as they make their way through evaluation. Or keep checking back here, to our blog, for updated information.




Get the latest discounts delivered straight to your inbox.


  • Adrian Rangel

    I gotta say i’m excited about this but also worried about the confusion it can cause and i believe the increase in phishing this can have. Not worried about me but people like my mom, which i tell her to always check the address bar if looks alike she will may be a victim of phishing

  • Prashant P Patil

    There should not be restriction on word basis. what if .car is some communities surname? will you stop them taking that TLD? Answer is no you can’t.