Capital is finite. Thinking that a gating network opposes basic venture capital prediction methods is a costly distraction.
Time is short to evaluate high-potential investments, and we must ration our efforts.
In reality, a gating network can refine these existing methods by integrating textual justifications with company attributes.
For example, using the outputs of a model based on LLMs as input for gating optimizes accuracy without wasting resources.
We cannot afford to waste liquidity in semantic debates.