1: \begin{proof} 

2:   Our proof is not rigorous. 

3: 

4:   %First, by Proposition \ref{prop:consensus}, we know that agents

5:   %always reach a consensus as $k$ becomes large (as $k>r$, to be

6:   %precise). Now, 

7:   Since agents are 

8:   homogenous with respect to tolerance $\eta$, they will all

9:   \emph{jointly} increase their weight to a 

10:   particular agent $j$ (or they will \emph{jointly} not do

11:   so). Therefore, as $k$ increases, rows of $\mathbf{W}^{(k)}$ become more

12:   and more similar, independent of the initial conditions

13:   $\mathbf{W}^{(1)}$ (if weight matrix $\mathbf{W}^{(1)}$ is identical

14:   in each row, this will propagate to any $\mathbf{W}^{(k)}$ with

15:   $k>1$, but even if not, rows will become more and more similar by

16:   the homogeneity of agents). The

17:   weight mass that any particular agent $i$ assigns to any particular

18:   agent $j$ is clearly proportional to $F_{j,k}(B_{k,\eta})$ (cf.\ Figure

19:   \ref{fig:intro}) since this value indicates how frequently agent $j$ is

20:   truthful. Hence, since rows of $\mathbf{W}^{(k)}$ are

21:   (approximately) identical, as $k$ becomes large, with each entry

22:   $[\mathbf{W}^{(k)}]_{ij}$ 

23:   being proportional to $F_{j,k}(B_{k,\eta})$, limiting beliefs of agents

24:   are given by,  

25:   \begin{align*}

26:     b_i^k(\infty) \approxeq b_i^k(1)=\sum_{j=1}^n\lambda_j b_j^k(0),

27:   \end{align*}

28:   where $\lambda_j\propto  F_{j,k}(B_{k,\eta})$. 

29:   This completes the proof. 

30: \end{proof}

31: