How to make almost anything

An MIT course on digital fabrication, “How to Make (Almost) Anything,” prompts debate over breadth‑first, hands‑on teaching versus deep specialization. Supporters argue that an intense, survey-style class that spans CAD, electronics, machining, and manufacturing gives students crucial “what’s possible” awareness and T‑shaped skills, especially valuable in an era of AI tools and maker spaces. Critics question the superficial treatment of complex disciplines and the limited usefulness of the public course materials for independent learners.

Course scope and intensity

  • Thread centers on MIT’s “How to Make (Almost) Anything” as a survey-style, fabrication-heavy course.
  • Officially ~18 hours/week; multiple commenters report real workloads closer to full‑time, with 40+ hours/week common.
  • Emphasis is on lab work: CAD, electronics, PCB design, CNC, machining, etc., with students building weekly assignments and a substantial final project.
  • Student project galleries and personal accounts describe the class as “life‑changing,” shifting some into applied engineering careers.
  • Unusually high staff ratio (many TAs) and follow‑on courses (e.g., “How to Make Something that Makes (Almost) Anything”) are noted.

Breadth vs. depth and educational philosophy

  • Critics: the site and syllabus look like long lists of tools/buzzwords; content seems too superficial for serious CAD or manufacturing; some even mock stylistic choices like all‑lowercase text.
  • Defenders: this is intentionally a breadth‑first “survey course,” meant to give cross‑disciplinary exposure and awareness of what’s possible, not job‑ready expertise.
  • Several argue that university is for broad conceptual foundations, not narrow vocational training, and that “making poorly” is a necessary and valuable stage.
  • Comparisons to makerspaces/TechShop: short tool trainings plus lots of independent practice; an overview that later enables deep specialization.
  • Discussion of “T‑shaped” or even more complex (“centipede‑shaped”) skill profiles and the rising importance of horizontal knowledge in the LLM era.

LLMs, CAD, and automation

  • Multiple comments describe using LLMs to generate STL files, write CAM/CNC workflows, or act as CAD “helpers”; perceived as surprisingly effective for small projects.
  • Others argue LLMs should be used to write reusable automation scripts rather than doing every task directly.
  • Limitations raised: API upload caps, performance for thousands of documents, and difficulty handling messy scanned documents (handwritten checks/invoices).
  • Broader meta‑discussion: LLMs have vast but latent knowledge; human generalists add value by knowing what to ask for, steering away from “obvious” but suboptimal solutions.

Manufacturing at scale

  • One commenter notes a gap between prototyping and at‑scale industrial production (e.g., Shenzhen experience).
  • Responses: this knowledge is hard to generalize; real understanding of DFM, factory selection, and automated QA usually comes only from going through production a few times.
  • Pointers given to manufacturing engineering programs and region‑specific guides.

Instructor controversy

  • Several comments link to documents about the instructor’s contact with Epstein and imply moral judgment.
  • Others push back, criticizing “guilt by association” and arguing that merely appearing in such documents doesn’t by itself prove wrongdoing.