The basic building block of translating any entirely unknown script is figuring out a way to figure out what sounds each individual character represents. The most common tactic is to identify proper nouns that correspond to places or people known from sources in other languages.

The easiest way to do that is to find a text with a known language and the unknown language presenting the same message. The most famous example of that is the Rosetta Stone, which enabled the first translations of Demotic Egyptian and Egyptian Hieroglyphs. Once the proper nouns are identified, the you use the the sounds of the words you definitely know to figure out what sound the unknown symbols make in those names and then apply those sounds to other words in the undeciphered script, working backwards to identify whole words and translating those based on the language you do know.

Another option is to work backwards from known languages hypothesized to be related to the unknown script. Prominent examples of this include Old Persian Cuneiform and Mycenean Greek Linear B. In both cases, the languages were ultimately cracked by identifying frequently repeated words and taking a stab in the dark that they were probably the names of people and places known from Classical Greek sources. In both cases, translators then applied the sounds they associated with those names to other words in the unknown languages to see if more recognizable words with cognates in known languages appeared. Once intelligible phrases emerged similar to other Iranian languages and Ancient Greek, they could apply known linguistic patterns to understand things like grammar and syntax.

The problem with the Indus Valley Civilization is that we don't have any clear references to their city names or rulers in other languages, nor do we have any multilingual texts with Indus Valley script as one of the languages represented. This means there is no starting point to even make educated guesses based on the theories that their language would be related to modern Dravidian languages or ancient Elamite. Without something to use as a starting point and identify what individual symbols might indicate, translators have nothing to work from.

With the available evidence today, there's not much hope that the Indus Valley script will ever be translated, but there is one potential avenue. Just a few months ago, Francois Desset published his research on translating the Linear Elamite script, used primarily in southwestern Iran from about 2300-1800 BCE. Linear Elamite actually does have a multi-lingual text, shared with Akkadian, and names were identified as early as 1905. However, it took more than a century for enough individual examples of Linear Elamite to surface for a sufficient number of words to be identified based on the known proper nouns. Once those texts were available, the systems described above were employed to translate the entire script.

Linear Elamite provides some interesting possibilities for the Indus Valley Script, but also the Proto-Elamite Script. Much like the Indus Valley script, Proto-Elamite has no multilingual texts and is either too old or structured too differently from Elamite Cuneiform to be translated purely on the basis of similarities to the known language. It doesn't help that early Elamite history is poorly documented, meaning we don't know many proper nouns to start making guesses with either. Proto-Elamite and the Indus Valley script appear to share a few symbols, and Linear Elamite appears to share symbols with both. If Linear Elamite can be used as a foundation, Proto-Elamite can probably be translated, and the shared signs in both of those Elamite scripts may be enough to give us a starting point on the Indus Valley Script, especially if those symbols can be used to identify any recognizable names.