how are people handling the discontinuity problem when forecasting star growth over a multi-year horizon?

the 2027 deadline is long enough that pure trajectory extrapolation feels like a shaky foundation for most nominations here. open-source adoption tends to move in jumps rather than smooth curves, a major framework picking up a library as a dependency, a competitor going closed or paid, a high-visibility tutorial or conference talk that hits at the right moment. those events can compress what would otherwise be years of organic growth into a short window, and they're genuinely hard to anticipate from baseline data alone. so curious whether people making calls on this list are primarily anchoring on current velocity and adjusting for known catalysts, or whether there's a more structured approach being used, something like analogy forecasting against projects that already crossed the 25k threshold under roughly comparable conditions. ecosystem positioning seems like it would matter a lot here too, since a library sitting at the intersection of two fast-moving areas probably has a different risk profile than one in a more stable niche, even if the raw trajectory looks similar at baseline. is there a particular signal or combination of signals that's actually doing the predictive work when you make a nomination, or is it mostly calibrated intuition with trajectory as a rough sanity check?