---
title: "Manas Vardhan (@manasvardhan) | Hashnode"
description: "I write philosophy, literature and tech. Read the latest articles by Manas Vardhan on Hashnode."
url: "https://hashnode.com/@manasvardhan"
---

@manasvardhan

# Manas Vardhan

@manasvardhan · Los Angeles, USA · Joined March 2026

I write philosophy, literature and tech

[Website](https://manasvardhan.github.io/)

### [The MCP Tax Is Real, and It Is Quietly Killing Your Agent's Reasoning](https://theagentstack.hashnode.dev/the-mcp-tax-is-real-and-it-is-quietly-killing-your-agent-s-reasoning?utm_source=hashnode&utm_medium=feed)

Every time your AI agent makes a tool call through MCP, it pays a tax. Not in dollars (though that too), but in tokens. Tens of thousands of them. Silently injected. Every single turn. A new paper dropped yesterday that puts hard numbers on this prob...

8 min read · [theagentstack.hashnode.dev](https://theagentstack.hashnode.dev/the-mcp-tax-is-real-and-it-is-quietly-killing-your-agent-s-reasoning?utm_source=hashnode&utm_medium=feed)

[1](https://hashnode.com/posts/the-mcp-tax-is-real-and-it-is-quietly-killing-your-agent-s-reasoning/69eb8de3f9c04bf5ebda8a33) [Apr 24](https://hashnode.com/posts/the-mcp-tax-is-real-and-it-is-quietly-killing-your-agent-s-reasoning/69eb8de3f9c04bf5ebda8a33)

### [RL in the Pre-train Space: Why Training on P(y) Beats Training on P(y|x)](https://theagentstack.hashnode.dev/rl-in-pretrain-space-training-py-beats-pyx?utm_source=hashnode&utm_medium=feed)

A new paper shows that reinforcement learning directly on the marginal distribution unlocks reasoning capabilities that standard RLVR can never reach.

7 min read · [theagentstack.hashnode.dev](https://theagentstack.hashnode.dev/rl-in-pretrain-space-training-py-beats-pyx?utm_source=hashnode&utm_medium=feed)

[Apr 16](https://hashnode.com/posts/rl-in-pretrain-space-training-py-beats-pyx/69e101636ef723c0870126bb)

### [The Three Walls Your AI Research Agent Keeps Hitting](https://theagentstack.hashnode.dev/the-three-walls-your-ai-research-agent-keeps-hitting?utm_source=hashnode&utm_medium=feed)

A Meta paper reveals the overfitting wall everyone accepted was actually evaluation noise, and the real ceiling is much further out

8 min read · [theagentstack.hashnode.dev](https://theagentstack.hashnode.dev/the-three-walls-your-ai-research-agent-keeps-hitting?utm_source=hashnode&utm_medium=feed)

[Apr 4](https://hashnode.com/posts/the-three-walls-your-ai-research-agent-keeps-hitting/69d06bbd7c88b94539041e69)

### [Chain-of-Thought Was Supposed to Be Our Window Into AI Reasoning. Optimization Is Slamming It Shut.](https://theagentstack.hashnode.dev/cot-monitorability-optimization-hides-reasoning?utm_source=hashnode&utm_medium=feed)

Here's the deal we thought we had with chain-of-thought prompting: let the model show its work, and we can watch the reasoning unfold. If something goes wrong, we'd see it in the chain. CoT was our audit trail, our interpretability shortcut, our free...

6 min read · [theagentstack.hashnode.dev](https://theagentstack.hashnode.dev/cot-monitorability-optimization-hides-reasoning?utm_source=hashnode&utm_medium=feed)

[Apr 2](https://hashnode.com/posts/cot-monitorability-optimization-hides-reasoning/69cec6985928c28d1e52de9f)

### [Tucker Attention: GQA, MLA, and MHA Were the Same Thing All Along](https://theagentstack.hashnode.dev/tucker-attention-gqa-mla-mha-same-thing-all-along?utm_source=hashnode&utm_medium=feed)

All major attention variants are special cases of one tensor decomposition, achieving 10x parameter reduction with zero performance loss

7 min read · [theagentstack.hashnode.dev](https://theagentstack.hashnode.dev/tucker-attention-gqa-mla-mha-same-thing-all-along?utm_source=hashnode&utm_medium=feed)

[Apr 2](https://hashnode.com/posts/tucker-attention-gqa-mla-mha-same-thing-all-along/69cec5402d02ee22cb0df18f)
