In Narva, the Estonian city pressed against the Russian border, the municipal administration plans to significantly reduce its participatory budget. In neighboring Narva-Jõesuu, there's talk of redirecting those resources to non-governmental organizations. It's worth saying from the outset: these are plans and considerations reported by ERR, Estonia's public broadcaster. Not a confirmed cancellation. The program is still under review. The reasons cited are the same ones that come up everywhere: neighborhood proposals that are costly, vague, or hard to execute.

A participatory budget is a process where residents propose and vote on how a portion of public money gets spent, because it aims to close the gap that representative democracy leaves between actual needs and what officials manage to see. That's its promise. The popular thesis now circulating in Narva, though more whispered than declared, holds that the promise has run its course. If residents ask for impossible things, better to let the professionals decide.

That argument deserves credit. Whoever runs a city answers for every euro. A proposal that says "more life in the neighborhood" can't be put out to tender, budgeted, or audited. If a project costs three times what was planned, or requires permits the municipality doesn't control, someone has to absorb the frustration. Usually it's the official who has to tell enthusiastic voters "this can't be done." That wear and tear is real. Stones Don't Lie talks about fatigue as the limit of any participatory structure. Human attention is finite. Records from Swiss referendums show the same people always participate, while everyone else effectively delegates.

The problem might not be that people propose badly. The channel might be receiving the signals badly. A vague idea isn't a citizen failure — it's an unfiltered signal. Participation works like a channel with bandwidth requirements, filtering, and processing capacity. When any suggestion comes in without help turning "we want a better park" into a project with scope, cost, and timeline, the result is noise. Then someone blames the input.

The loop breaks. Citizen input should translate into an executable output, and feedback about what was done and why should go back to whoever proposed it. When technical support is missing, or feasibility criteria aren't explained upfront, the process fails. A broken process invites shutting it down rather than fixing it. I don't know exactly which failure is occurring in Narva. The sources don't detail it. That's why the question matters more than the verdict.

This gap between voice and execution goes back to the Athenian agora. Any citizen could speak. What actually came up for a vote passed through a council that prepared the agenda, and what got decided needed magistrates to carry it out. Voice without an administrative channel was just noise in the square. Whoever filtered the proposals held enormous power.

Porto Alegre in Brazil shows the other side of the coin. Participatory budgeting began there in 1989 as a management experiment and became an international model. What sustained it were clear rules and technical support from the municipal government. In several places where it was copied, it decayed once it lost political backing. Kerala in India shows similar light and shadow — notable results alongside capture by local elites and citizens' difficulty evaluating complex projects. Participation holds up when someone pays for the work of translating it.

Now, cui bono — because no budgetary decision is neutral. If Narva cuts, the administration gains control. Less money decided by residents means more money decided from the office. If Narva-Jõesuu redirects funds to NGOs, those organizations would gain funding. The incentive is there. In both cases, a convenient narrative advances for governments that prefer deciding from the top down: that participation is inefficient. It's a good story because it's partly true. That's how good stories work. Debating which structure is better is easy. The hard part is knowing who holds the pen.

Redirecting to NGOs doesn't solve the bottleneck either. An NGO may have more technical capacity than a group of neighbors. But then it's no longer the resident deciding — it's whoever has the capacity to write up projects. Swapping one filter for another doesn't improve the channel.

It's worth experimenting before cutting. A year with three components: a technical team that helps turn ideas into projects before the vote, feasibility criteria published from the start, and mandatory feedback for every rejected proposal, with reasons and possible adjustments. Then measure how many proposals actually get executed. If the percentage stays low even with that support, the cut would have a solid argument behind it. If it rises, the original diagnosis was wrong from the start.

There's a detail that contradicts intuition. Most people assume a participatory budget dies when people stop participating. What usually happens first is the opposite. People keep proposing. It's the administration that loses capacity or will to process what it receives. Vague ideas. Limited resources. Accumulated frustration.

What if the real limit isn't in residents' proposals, but in our ability to give them shape?