About Me

Quick Logistics

• Lectures: Tue & Thu, 2:00–3:15 PM, TD-W 1701 (Sanjay covers when Arpit is traveling)
• Discussion: Fri, 1:00–1:50 & 2:00–2:50 PM, NH 1109 — attendance required
• Communication: Slack — join today
• Grading: PAs (40%), Midterm (20%), Project (20%), Participation (20%)
• Participation: Discussion attendance + PA oral check-ins (can you explain your own code?)
• Midterm: May 5, in-class, closed-device — no LLMs, no notes
• Book: First Principles of Networking — free, online, work in progress
• Full details: course website

Your course team — all SNL PhD students building the tools you’ll use:

Sanjay Chandrasekaran (TA)	Sylee Beltiukov (TA)	Jaber Daneshamooz (Staff)	Manni Moghimi (Staff)

My North Star: Performant Internet for All

The Internet transformed work, healthcare, education, and government. But connectivity remains deeply unequal:

My research focuses on two things: building systems that make high-quality connectivity cheaper to deliver, and building the data infrastructure to measure whether we’re succeeding.

The Data Gap: What We Can’t See, We Can’t Fix

Policymakers allocate billions based on ISP self-reported data.

The problem: ISPs have incentives to overstate coverage and quality.

• The FCC’s National Broadband Map relied on ISP certifications — nobody independently verified them
• Crowdsourced speed tests (Ookla, M-Lab) lack context: a “slow” result might be a cheap plan working correctly, or a premium plan failing
• No independent dataset for broadband affordability existed at all

“What we can’t see, we can’t fix.” — Benton Institute OpEd, January 2026

BQT: Independent Broadband Data at Street-Address Granularity

We built BQT — it automates querying ISP websites to extract advertised speeds and prices at individual street addresses.

BQT enabled querying 1M+ addresses for 50+ ISPs, bridging three critical data gaps:

What BQT Revealed: A $10 Billion Audit

The Connect America Fund (2011–2021) spent $10 billion subsidizing ISPs to serve high-cost rural areas.

We used BQT to audit what was actually delivered (SIGCOMM ’24):

BQT’s Impact

One research tool — bridging the gap between measurement and accountability at every level of government.

What Lies Ahead: From Auditing to Continuous Monitoring

BQT is now shaping legislative decisions across the country:

• California: CPUC using BQT data to audit ISP compliance and support digital discrimination cases
• Virginia: JCOTS broadband affordability report delivered to the state legislature
• New York City: Partnership with NYC Office of Technology & Innovation on broadband equity assessment

Next: BQT is positioned to track the $42B BEAD program — independent accountability at national scale.

And we’re building NetVibe — a continuous broadband quality monitoring platform that captures application-level quality of experience. We’ll discuss NetVibe later in the course.

Path 2: Self-Driving Networks

The second research path: making networks that operate themselves.

Why this matters for my north star: Community networks can’t afford large operations teams. Self-driving networks bring down the cost of delivering performant connectivity.

Why this matters for the AI era: This is the networking instance of a broader challenge — embedding ML models inside reliable, trustworthy operational loops.

We are working with Google and the Department of Energy to build the representation layer that enables agentic systems to manage complex, massive networks — with minimal human intervention.

From Research to Tools: The Data Generation Ecosystem

The self-driving networks pipeline required a data generation substrate — infrastructure for creating realistic, repeatable network experiments at scale.

That substrate became the tools you’ll use this quarter:

These aren’t textbook exercises. They’re research infrastructure built by my PhD students — the same TAs who teach your discussion sections.

What My Research Taught Me

Every breakthrough came from asking the right first question — not from knowing the right mechanism.

Computing Generates the Future

Computing is a generative discipline — it produces the abstractions from which entirely new problem spaces emerge.

Each row: a specific computing abstraction that opened a problem space with its own theory, methods, and communities.

AI Is a Product of These Layers

AI emerged from decades of computing abstractions stacked on top of each other:

Packet switching → IP thin waist → HTTP/Web → Cloud/virtualization → GPU clusters + distributed training → LLMs

The Frontier Has Moved: From LLMs to Agentic Systems

Two years ago: LLMs will subsume everything.

Today: agentic systems — AI models embedded in larger frameworks that plan, use tools, maintain state, and interact with environments.

The shift happened in under two years. It demanded entirely different foundational knowledge.

What This Means for Your Education

Whatever follows agentic systems will require a different combination of computing abstractions.

The value of your education: learning to reason at the level of the abstractions that generate paradigms.

“Students who understood decomposition, resource allocation, and formal specification navigated the paradigm shift — because they were operating at the level of the abstractions that generate paradigms.”

This course teaches that level of reasoning — applied to networked systems.

First-Principles Thinking + AI = Superpowered Engineers

An LLM can recite TCP, describe AIMD, list 802.11 frame fields. That recall is now automated.

Three Capabilities You’ll Build

By the end of CS 176C, you will be able to:

1. Analyze — Given any networked system, identify its anchor constraint and trace how it shapes every invariant answer

2. Predict — When a constraint shifts, predict how the system must restructure — before looking at what engineers actually did

3. Design — Reason through how to build a new system under novel constraints, using the same framework

These capabilities are your leverage for AI collaboration.

Puzzles This Course Will Solve

March 2020: The EU asks Netflix to reduce streaming quality to prevent network collapse. Why did that help — and which three systems are coupled in the dependency chain?

2020: Zoom surges with remote work — and chooses UDP instead of TCP. What constraint forces that choice?

2025: 5G radio delivers 1ms latency. So why does your phone still buffer?

You already know these events happened. This course teaches you why they had to happen — by tracing the structural constraints that forced each outcome.

The Core Idea

Every networked system must answer four structural questions:

The answers differ radically across systems. The questions never change.

The Anchor

Every system has one constraint that shapes all four answers.

Find the anchor → the design becomes predictable.

Example: Loading a Web Page

You open your laptop, connect to WiFi, and type www.google.com. What happens?

Visualizing the Web Page Load

DHCP and ARP are local — confined to the LAN. DNS, TCP, and HTTP traverse the Internet to reach remote servers. Each protocol is color-coded to show the dependency chain.

Three Protocols, Three Design Patterns

DHCP centralizes — unique addresses need one authority. TCP distributes — decentralization prohibits a central scheduler. DNS delegates — namespace too large for one server, too structured for full distribution.

Same four questions. Radically different answers. Each forced by different constraints.

The Book & What’s Next

Welcome Aboard

Before Thursday: (1) Join Slack, (2) Read Ch 2: First Principles, (3) PA1 drops Wednesday, (4) Check the course website

For curious minds:

• Computing Is a Generative Discipline
• Systems for Agents, Agents for Systems
• What We Can’t See, We Can’t Fix

Questions?