Systems: 7 ideas from MIT courses

Most overruns trace back to missing or over-ambitious requirements. Write what the system must do, measurably, and test against it. The Mars Climate Orbiter was lost when a units rule never reached the build.

In practice: Match rigor to the stakes. The DC-3 was specified in 3 pages and a phone call.

Fundamentals of Systems Engineering · MIT 16.842 · fall 2015

Requirements say what to achieve, and how well. Specs say how it gets built. Owners jump to a tool (“we need a chatbot”) before naming the outcome.

In practice: You sign off on the outcome, so the tool can change later. Don’t let a named tool become the requirement.

Fundamentals of Systems Engineering · MIT 16.842 · fall 2015

Teams polish the parts and neglect the hand-offs between them. The Ariane 501 rocket was lost when one part’s test data was read as flight data. In a business, leads die quietly in the gaps between tools.

In practice: The fix is cheap: name every hand-off, define what crosses it, test it with real data.

Fundamentals of Systems Engineering · MIT 16.842 · fall 2015

Verification asks: was it built right, to the written requirements? Validation asks: was the right thing built? Every metric can be green while the business still books no jobs.

In practice: Verification can be automated. Validation needs a real signal, like a booked call or a paying customer. It is the check owners skip.

Fundamentals of Systems Engineering · MIT 16.842 · fall 2015

Launch starts the longest phase: running, watching, fixing and upgrading. A machine nobody watches decays as the market, the ad platforms and the audience shift.

In practice: This is the engineering case for a monthly relationship over a one-time build. Decide at design time who runs it, and how often.

Fundamentals of Systems Engineering · MIT 16.842 · fall 2015

Capacity, process, sourcing, people and software must fit the plan and each other. A great ad account and a great closer still lose if the ads promise what the process can’t deliver.

In practice: Run the fit test before you add a tool or channel, and again as markets move.

Operations Strategy · MIT 15.769 · fall 2010

System dynamics studies feedback loops, stocks and flows. A metric is an output, so chasing it treats the symptom. A small change in the loop compounds.

In practice: Loops are easy to draw and hard to time. Confirm the real delay before you act, or you’ll overcorrect.

Introduction to System Dynamics · MIT 15.871 · fall 2013 (Profs. John Sterman & Hazhir Rahmandad)