Is there a reference for a randomized communication lower bound for the following problem: $\textsf{Or-of-Equalities} : \{0,1\}^{n^2} \times \{0,1\}^{n^2} \to \{0,1\}$ defined by $f(x,y)=1$ iff there is some $i \in [n]$ s.t. $x_{(i-1)n+1}\cdots x_{in} = y_{(i-1)n+1} \cdots y_{in}$? The "unparallelized" version is simply the $\textsf{Equality}$ function on $n$ bits, which has $\Theta(\log n)$ randomized communication complexity. I would hope that the bound is something like $\Omega(n)$ :)
For context, I was thinking about the $\textsf{Sink}$ function used by Chattopadhyay, Mande, and Sherif to disprove the log-approximate-rank conjecture [J. ACM '20]. This function is almost the same, except each pair of equalities overlaps in a single input coordinate (for which the required value differs between the two equalities). They get an $\Omega(n)$(-ish?) bound using the corruption method, which may also go through for this other function --- I'm not sure. The function also resembles $\textsf{Tribes}$ - except when you look up randomized communication complexity of $\textsf{Tribes}$, the results (e.g. Jayram, Kumar, and Sivakumar, STOC '03) lift using $\wedge$, not $\oplus$ (a.k.a. equality).
Thanks!
Noah