The intractability of decoding generic linear codes is at the core of an important branch of post-quantum cryptography.
In this context, the code is random by design or it is assumed to be so in the security reduction.
This talk will focus on versions of the Decoding Problem where the error vector is structured, in general to achieve better performance.
While combinatorial techniques such as Information Set Decoding are often the method of choice to attack these versions, I will describe the potential of algebraic algorithms.
I will mostly consider the Regular Syndrome Decoding Problem and a paper presented at Eurocrypt 2023.
I will also mention ongoing work on an assumption used in the CROSS submission to new call for signature schemes launched by NIST.