What I find interesting is that biotech and semiconductor have barriers high enough that you need to be high quality to even think of attempting. But lower bar

What I find interesting is that biotech and semiconductor have barriers high enough that you need to be high quality to even think of attempting. But lower barrier areas like web and software have a lot more people trying but similar numbers as biotech getting funded. This to me means that the hard stuff needs lower barriers because a larger sample size should be correlated with a wider variety of ideas. More stupid ideas sure, but also more of ideas so stupid they just might work. The numbers funded is strange. Take other for example. Are the ideas in Other really that bad? Why are the numbers funded all within 2 magnitudes of each other regardless of number of people trying? My guess is that active VCs form a bottleneck on ideas and the numbers funded is more strongly coupled to the number of active VCs in the area than the average merit of the ideas. And that the number of VC per area does not vary too much, with areas like software and web having the most participants.