This thread is for sharing some perspective on the role of projections. I have received many notes that our previous projections turned out to be true. This actually defeats the purpose of our work. Modelers project so that policymakers can gauge the future and take actions. 1/n

Modelers do NOT want to see their projections validated by data for this particular context. The first sign of the uptick was around mid-February when India was reporting 11000 cases and 91 deaths. Taking stringent measures is hard when the numbers seem small and manageable. 2/n

But, when you see a consistent trend, you have to intervene, relying on the projections of how bad it could get under plausible scenarios. This is THE treacherous thing with this virus. You have to anticipate its silent footsteps. You cannot wait to see rising deaths to act. 3/n

We need to track the data real time, picking up such clues, across states, and intensify public health interventions nimbly. Once the healthcare capacity is overwhelmed deaths start rising, not because the virus is clinically more lethal, but people cannot get care they need. 4/n

We cannot change the staggering numbers that we are seeing right now but we can change the course of this pandemic and try to beat the dire projections. That is the real success of a modeling effort, not to see the projections come true but drive actions so that they DO NOT. 5/n

I feel frustrated that we have spent 9 weeks just updating projections when we all know what exponential growth is like. Modeling is useless when it cannot inform actions. Large populous states like UP, WB, Bihar with high R values are ticking time bombs. We need actions. 6/n

All models are saying qualitatively the same thing. We can argue when the peak is & how high the peak is but the reality on the ground tells us much more than models at this stage. We cannot wait any longer to validate projections by real data. The real data are real people. n/n