You seriously self-study an MIT OpenCourseWare course by choosing one that ships problem sets and exams with solutions, giving it a fixed term you set yourself, and treating the problem sets as the course. Our recommendation is about 8 to 10 hours a week for about 18 weeks, a little slower than the roughly 12 hours a week an MIT student puts into a full class. The lectures are there to help you do the work, and the timed exams at the end are how you prove you did it.
Quick Answer: how do you self-study an MIT OCW course for real?
Pick an OCW Scholar course, marked SC in the number, or any course with problem set and exam solutions. Check the prerequisites, put start and end dates in your calendar, and give it about 8 to 10 hours a week. Do every problem set before checking solutions, then take the exams timed and closed-book.
Why people stall on MIT OpenCourseWare
People stall on OCW because it removes everything that makes an MIT student finish: there is no enrollment, no start or end date, no grading and no credit. You get the materials and nothing else. Without a deadline or a grade, watching lectures feels like progress while the problem sets, the real course, never get done.
Start with what OCW actually is. Since 2001, MIT OpenCourseWare has shared MIT course materials openly, beginning with a pilot website of 50 courses and an official launch in 2003 with 500. Today the OCW about page lists 2500+ MIT courses and 500 million+ learners and educators, under a Creative Commons license. It is a library, not a school, and the page is clear about the trade. "MIT does not offer credit or certification to users of OCW." On timing, it promises "No enrollment and always available," with no start or end dates. A course that is always available is a course you can always start next week.
The second trap is the format people meet first. The MIT OpenCourseWare YouTube channel has over 6M subscribers, and for most learners a lecture video is the front door. But the 18.06 Linear Algebra syllabus says it plainly: "The homeworks are essential in learning linear algebra." A playlist does not hold the homework. If you only watch, you have seen the course, not taken it.
The third trap is picking a course you are not ready for. MIT lists prerequisites for a reason. The 6.006 Introduction to Algorithms syllabus says, "We strongly caution against taking 6.006 before having fulfilled the listed prerequisites." MIT students even submit a short Problem Set 0 that checks them. At home, nobody stops you, so you find out in week three.
The last trap is pace. MIT's Registrar defines a unit as "approximately one hour of effort per week, or 14 hours per term, for a typical full-term subject," and the Registrar's standard example is a 12-unit subject. That is about 12 hours a week, roughly 168 hours over a term. Most self-learners budget two or three hours, fall behind, and decide the material is too hard when the real problem was the schedule.
How to pick the right OCW course
Pick a course you can grade yourself on. That means an OCW Scholar course, or any course whose resource list includes problem sets with solutions and exams with solutions. Then check the prerequisites on the syllabus and be honest about them. A good course you are ready for beats a famous one you are not.
Start with the Scholar list. The OCW Scholar course list describes these as "complete introductions to essential math, science, engineering and economics subjects, designed especially for independent learners." They carry SC in the course number, such as 18.06SC and 18.01SC.
Check the resource types before you commit. Every OCW course page lists what it includes. You are looking for two things: problem sets with solutions, and exams with solutions. Without solutions you cannot check your work, and without exams you cannot prove you learned it. 18.06SC Linear Algebra is the model here. It has lecture videos by Gilbert Strang, summary notes for every video with suggested readings in his textbook, problem-solving videos on every topic by an MIT recitation instructor, problem sets with solutions, Java demonstrations, and a full set of exams with solutions and review material. 18.01SC Single Variable Calculus has the same shape: problem sets and solutions, exams and solutions, and problem-solving videos, running from differentiation and integration through to infinite series.
Know what a course is missing. 6.0001 Introduction to Computer Science and Programming in Python is a strong start for people with little or no programming experience, but its listed resources are lecture notes, lecture videos, problem sets and programming assignments with examples. No problem set solutions are listed. You can still take it, but you will need another way to check your code, such as tests you write yourself or a study partner who reviews it.
Read the prerequisites line. The 18.06 Spring 2010 syllabus lists Multivariable Calculus (18.02) as a prerequisite. 6.006 asks for 6.0001 and 6.042J Mathematics for Computer Science, which covers discrete math such as sets, logic, proofs, recursion, graph theory and probability. If you are missing a prerequisite, take that course first. That is not a detour, it is the route.
Your realistic options, compared
There are five realistic ways to work through MIT material on your own, and they differ mainly in whether anything checks your work or sets a deadline. The free OCW course page is the most complete. The others add grading, a credential, or people, each with its own cost.
| Option | Cost | Graded work | Deadlines | Credential | Best for |
|---|---|---|---|---|---|
| OCW course page alone | Free | None; you grade yourself with posted solutions | None; you set them | None; MIT gives no credit or certification for OCW | Self-starters who will keep their own term |
| OCW YouTube playlist only | Free | None | None | None | A first look at a subject, not learning it; the problem sets are missing |
| MITx on edX, audit or verified | Audit is free with temporary access; verified is paid | Graded assessments on the paid verified track only | Depends on the course; audit access is temporary | A certificate on the paid verified track | People who want graded work and a certificate |
| A study group or accountability partner | Free | Each other's work, informally | Set by the people in it | None | People who keep promises to others better than to themselves |
| LearnPath | Free plan is step 1 of a path, for 2 paths (some new accounts are asked to start a 7-day Pro trial, card needed, before their first lesson); Pro $12.99/month or $8.99/month billed annually | A quiz after each lesson; not MIT's problem sets | None built in | Pro-only LearnPath certificate; not accredited and not MIT credit | People who want YouTube lessons put in order and checked |
The honest summary is that the OCW course page is enough if you keep a schedule. It has the problem sets, the exams and the solutions to grade yourself. What it lacks is pressure. A playlist alone is not a course at all.
edX explains that auditing is free and gives you the materials for an audit period, so access is temporary. To take graded assessments and get a certificate, you enroll in the paid verified track. MITx courses run on edX, so that is the route if you want MIT material with outside grading.
LearnPath is one row in this table, not the answer to it. It builds a path from YouTube lessons on any topic, in order, with a quiz after each lesson. The Free plan gives you step 1 of each path, for 2 paths, though some new accounts are asked to start a 7-day free Pro trial (card needed, $0 today) before their first lesson. Every later step, unlimited paths and the certificate are on Pro. Those quizzes are not MIT's problem sets, and the certificate is not MIT credit. If you want a free credential tied to finishing a course, our guide on how to actually finish CS50 covers a course that gives a free certificate for passing every problem set and the final project at 70% or more.
An 18-week plan for 18.06SC Linear Algebra
This is our recommended plan: 18.06SC at about 8 to 10 hours a week for 18 weeks, which lands close to MIT's own budget of roughly 168 hours for a 12-unit subject, spread thinner. Every week runs the same loop: lecture, problem set, then solutions. The three unit exams and the final are built in as their own weeks.
We picked 18.06SC because it has everything a self-learner needs to grade themselves. If you are a programmer starting from Python instead, the same structure works for 6.0001 (or the newer 6.100L), with 6.0001's 6 problem sets and two closed-book quizzes standing in for the 18.06SC exams. Since 6.0001 lists no problem set solutions, test your code harder and get it reviewed.
Week 0: set your term and check you are ready
Write a start date and an end date 18 weeks later in your calendar. That is your term. Then check the prerequisite: the 18.06 Spring 2010 syllabus lists Multivariable Calculus (18.02). If calculus is rusty, start with 18.01SC instead. Our guide to learning the math for machine learning from YouTube explains why linear algebra pays off and where it fits. Get the textbook if you can: Strang's Introduction to Linear Algebra is the course text, and the summary notes point into it.
Weeks 1 to 4: Unit I and the weekly loop
18.06SC is split into three units, and each ends with a review and an exam: Unit I is Ax = b and the Four Subspaces, Unit II is Least Squares, Determinants and Eigenvalues, and Unit III is Positive Definite Matrices and Applications. Then comes a final course review and the final exam.
Each week, watch the lecture videos for that week's topics, read the summary notes, and do the suggested readings. Then do the problem set without looking at anything. Only after a full attempt, watch the problem-solving video on the topic and check your answers against the posted solutions. Mark every problem right, wrong, or right for the wrong reason. If you prefer the full lecture series in one place, Strang's 18.06 lectures from Spring 2005 are on the MIT OpenCourseWare YouTube channel as a playlist, starting with 1. The Geometry of Linear Equations.
Week 5: Exam 1
Stop new material. Redo every problem you marked wrong, cold, and work the Exam 1 review. Then take Exam 1 timed, closed-book, in one sitting, and grade it strictly with the official solutions.
Weeks 6 to 9: Unit II, the middle stretch
Return to the weekly loop. This is where self-learners stall, because the novelty is gone and the end is far away. Protect fixed time slots, not "when I have time". Falling a week behind is fine. Watching ahead without doing the problems is not.
Week 10: Exam 2
Same routine as week 5: the Exam 2 review, then Exam 2 timed and graded. Write down the two or three topics where you lost the most points.
Weeks 11 to 14: Unit III
Work through the last unit with the same loop. Spend any spare time on your weak topics, and do not add new resources now.
Week 15: Exam 3
The Exam 3 review, then Exam 3 under the same conditions as the first two. Add its weak topics to your list for the final.
Weeks 16 to 18: final review and final exam
Redo the hardest problem from every week in your log and work the final course review. Then take the final exam timed, closed-book, no phone, one sitting, and grade it against the official solutions. That score ends your term.
What to skip
Skipping the right things keeps an 18-week plan at 18 weeks. Skip lecture binges, rewatching whole lectures, peeking at solutions early, collecting extra resources, and courses you are not ready for. None of them moves you closer to a passing exam score, and all of them feel like progress.
Skip watching ahead. A lecture watched without doing its problems is mostly gone by the exam.
Skip full rewatches. If a problem stumps you, find the few minutes of lecture or problem-solving video that cover it, not the whole hour. Rereading and rewatching feel productive but test nothing, and we explain why in active recall vs rereading notes.
Skip early solution peeking. A solution read before a real attempt teaches you to recognize answers, not to produce them. Give every problem an honest try first, then check.
Skip resource hoarding. Extra textbooks and playlists are a way to avoid the one course in front of you. The Scholar course already has notes, readings and solutions.
Skip courses above your level. Jumping into 6.006 without 6.0001 and 6.042J is exactly what its syllabus cautions against. If you want discrete math first, the 6.042J Mathematics for Computer Science playlist is the place to start, paired with problem sets from its course page.
How to know you are on track
You are on track if every week ends with a finished problem set checked against the solutions, a log entry, and the next week's hours already in your calendar. Exam scores are the bigger checkpoints. If you are no more than one week behind and your exam grades are honest, you are doing it right.
Keep a simple log, one line per week: date, hours, problems attempted, problems right on the first try, and the idea that finally clicked. It shows when you drift, which topics cost points, and it becomes your record at the end.
Use these checks at the end of each week:
- The problem set is done before the solutions are opened. Every problem has a real attempt.
- You are at most one week behind. Two weeks behind means cutting extras, like optional readings, until you catch up.
- You can explain your wrong answers. If you cannot say why a solution works, redo the problem in a few days.
- Your exam dates have not moved. Moving an exam is how a term with an end date turns back into a course with no end.
Proof at the end
MIT will not give you anything for finishing an OCW course, so the proof has to be yours. The best proof is every exam in the course taken timed and closed-book, graded strictly with the official solutions, plus a log of every problem set. That is evidence you can show anyone, including yourself.
Keep the graded exams and your log together. If someone asks what you know, you can show exactly which problems you solved and how you scored.
If you want an outside credential, the options are different products. MITx on edX offers graded assessments and a certificate on the paid verified track. LearnPath's certificate is on Pro and is not accredited and not MIT credit. Neither replaces having done MIT's own problem sets and exams.
Frequently Asked Questions
These are the questions people ask most before they commit to an OCW course: how to start, how long it takes, whether there is a certificate, what Scholar means, whether the videos are enough, and where a beginner should begin. Each answer stands on its own, so you can jump straight to yours.
How do I seriously self-study an MIT OpenCourseWare course?
Pick an OCW Scholar course, or any course that posts problem sets and exams with solutions. Check the prerequisites, set your own start and end dates, and give it about 8 to 10 hours a week. Treat the problem sets as the course, attempt every one before checking, and take the exams timed and closed-book.
How long does an MIT OpenCourseWare course take to self-study?
MIT counts one unit as about one hour of effort per week, and a typical full class is 12 units, so a student spends about 12 hours a week, or roughly 168 hours over a term. At our recommended 8 to 10 hours a week, plan on about 18 weeks for one full course.
Do you get a certificate or credit from MIT OpenCourseWare?
No. MIT's own OCW about page says MIT does not offer credit or certification to users of OCW. The materials are free, open and always available, with no enrollment. If you need graded work and a certificate, look at MITx courses on edX, where the paid verified track adds graded assessments and a certificate.
What is an OCW Scholar course?
OCW Scholar courses are complete introductions to essential math, science, engineering and economics subjects, designed especially for independent learners. You can spot them by the SC in the course number, such as 18.06SC Linear Algebra or 18.01SC Single Variable Calculus. Both of those bundle problem-solving videos, problem sets with solutions and exams with solutions.
Is watching the MIT lectures on YouTube enough to learn the course?
No. The YouTube playlists hold the lectures, but not the problem sets, solutions or exams, and those are where the learning happens. The 18.06 syllabus says the homeworks are essential in learning linear algebra. Use the videos as support, and do the course work from the OCW course page.
Which MIT OpenCourseWare course should a beginner start with?
For math, start with a Scholar course such as 18.01SC Single Variable Calculus or 18.06SC Linear Algebra. For programming with little or no experience, 6.0001 or the newer 6.100L are introductions in Python. Do not start with 6.006 Introduction to Algorithms, because MIT strongly cautions against taking it before its prerequisites.
The bottom line
An MIT OpenCourseWare course is a full MIT course with the school taken out. You have to put back the term, the deadlines and the grading yourself. Pick a Scholar course, check the prerequisites, give it about 8 to 10 hours a week for 18 weeks, and do every problem set before you look at a solution.
Set your term dates tonight, and if you want lessons on another topic put in order, build a path that ends in a quiz you can pass before the term you set yourself is over.
