Why must a voting backend store its counts on the server — in a file or database — rather than in a JavaScript variable or in the browser?

Because the tally must be shared by all clients and survive restarts: browser storage is per-user, and an in-memory variable dies with the server process — only persistent server-side storage gives one durable, shared count.

* Only persistent server-side storage is both shared by all clients and durable across restarts. *

There are three places the count could live, and two of them fail:

• In the browser — each visitor would hold their own private copy, so no shared result could ever exist; worse, users control their browser and could set the count to anything.
• In a server variable (let counter = 0 next to the handler) — this is shared between clients while the program runs, since every request hits the same process. But it lives in RAM: any crash, restart, or redeploy silently resets every vote to zero.
• In persistent server-side storage — e.g. a votes.json file the handler reads with fs.readFileSync() and rewrites with fs.writeFileSync() (checking fs.existsSync() on first use). All clients update the same stored tally, and it survives restarts.

The file-system version is the simplest possible illustration — real applications almost always use a database instead — but the principle is identical: results must be stored on the server, in something that outlives both the browser session and the server process.

Go deeper:

• Node.js: File system (fs) API — existsSync, readFileSync, writeFileSync and their async counterparts.