The problem here isn't the data. It's the assumption that smoothing a line means forcing the chart to ignore what the data is actually doing. The user's scatter graph wobbles because duplicate values create flat stretches, then abrupt jumps. Removing duplicates and toggling hidden-cell settings doesn't fix the underlying issue: the chart is still trying to connect points that don't follow a clean, continuous sequence. That's not a glitch. That's the tool faithfully reflecting the structure of the input.
What this user is really asking for is a way to fabricate a curve that doesn't exist in the original dataset. They've already tried moving averages and found them lacking. They've considered generating new numbers but don't know how to produce values that would fit the curve. That instinct is worth pausing on, because it points to a deeper truth: a scatter plot with smooth lines is a visual approximation, not a statement of fact. If the data repeats, the line will show that repetition. If the data jumps, the line will jump. No chart setting will change that without altering the data itself.
The practical takeaway is to stop treating the chart as the problem and start treating the data preparation as the problem. If the goal is a smooth visual, the user needs to decide whether they're visualizing raw measurements or a modeled trend. Those are different tasks. A moving average trendline was a step in the right direction, but it failed because it was applied as a band-aid rather than as a deliberate modeling choice. The better move is to fit a curve to the underlying relationship using a method that respects the data's structure, then plot that fitted curve alongside the original points. That gives clarity without pretending the raw data is smoother than it is.
This is where the opportunity lies. Instead of hunting for a checkbox that will magically iron out the wobble, users should embrace the fact that spreadsheets are not drawing tools. They're calculation engines. The smooth line they want isn't hiding in a settings menu. It's waiting to be derived from the data itself. That means learning to use functions that generate fitted values, or stepping up to a tool that handles curve fitting natively. The user's instinct to generate new numbers was correct. The missing piece is knowing that those numbers should come from a model, not from guesswork. That's the real fix. And it's a skill worth building, because every wobbly chart you encounter is just a prompt to rethink the data, not the graph.