TutorFlow Simulations is live for every classroom on TutorFlow today.
A simulation here is a web page that runs a real calculation and redraws itself as a learner changes the values. You get one by writing a sentence about what you want them to work out. TutorFlow plans the model, builds the page, drives every control to check that the picture responds, and hands you a link. You never open an editor or write a line of code.
Explain a principle in words and learners write it down. Let them change the numbers and they work it out. Letting them change the numbers used to mean finding a developer, so mostly it did not happen.
Why the Simulation Never Gets Built
There are good simulation libraries on the web, and every teacher who needs one has already searched them. They cover the topics somebody funded. That is a real list, and it is never quite your list. It has a projectile launcher and no break-even model. It has an atom and nothing on how a sort behaves as the array grows.
A general assistant will write you a one-off page, and sometimes that page is fine. The trouble is that it writes everything from scratch every time. The slider markup, the animation loop and the axis labels are reinvented on each attempt, so the quality swings from one try to the next. You find out which kind of attempt you got while a class is watching.
Hiring a developer costs weeks and a budget nobody has for a single lesson.
So the lesson goes back to a slide with a graph on it, and the graph does not move.
How One Sentence Becomes a Working Model
TutorFlow starts where a teacher starts, with a sentence about the thing learners keep getting wrong.
"Simple against compound interest, so learners can see where the two curves part."
From that, TutorFlow writes a plan and stops. The plan takes a minute to read and it is yours to correct:
- The quantities it will track, and the rule that moves them
- Every control with its range, its starting value, and what moving it does
- The questions a learner has to answer by moving the controls, not by reading the screen
Reading the plan costs one credit. An idea that is wrong for your class gets rejected there, before anything expensive happens. Only when the model is right do you press build.
Every Control Is Tested Before You See It
A simulation whose numbers change while the drawing stays still is worse than no simulation. It passes every static check anyone would think to write. It looks right in a screenshot. It teaches nothing. That failure is the reason this step exists.
So nothing reaches you until TutorFlow has run the page and used it:
- Every control is driven from one end of its range to the other, and the screen is watched while it moves
- A version whose picture does not respond is rebuilt, not published
- Starting values are chosen so the effect shows the moment the page opens
The other half of the argument is what TutorFlow does not generate. The control panel, the fixed timestep loop, the reset, the formula and the question list are ours, and they are identical in every simulation. Only the model and the drawing are written fresh, which is the part that should differ.
A build takes a few minutes and you can leave the page while it runs. Editing happens afterwards in the chat beside the simulation. Write "widen the rate range to 20%" or "add an average line to the chart" and the next version has it. Every version is kept, so a change that makes things worse is one click from being undone.
Three Simulations You Can Open Right Now
Each of these started as one sentence and was built the way yours will be. They happen to be physics, finance and computer science. The shape fits any lesson where the relationship between two numbers is the point: break-even analysis, dosage curves, population growth, compound risk, algorithm complexity. Open one and change the values exactly as a learner would.
"A mass on a spring, for finding what the period does and does not depend on."
The period of a mass on a spring hangs a block on a spring and plots its displacement against time. Take the mass down to 0.4 kg and the block hurries. Soften the spring to 5 N/m and it slows again. The measured period settles at 1.78 seconds, which is what the formula printed beside it gives for those two numbers. Nobody typed that in.
"Simple against compound interest, for seeing where the two curves part."
Simple vs. compound interest draws both on the same axes and shades the gap between them. At 8% over ten years on $1,000 the gap is $358.92, which is small enough that the two lines still read as one. Push the rate to 20% and the term to 30 years and the argument makes itself.
"Three sorting algorithms counted against each other as the array grows."
How many comparisons a sort makes shows the array as bars with the pair under comparison picked out, and plots the running total against array size underneath. Grow the array and the curve bends. Switch the input from reversed to nearly sorted and insertion sort's curve flattens, which is the answer to a question that is hard to make convincing on a whiteboard.
What a Simulation Costs
Simulations use the same credits as the rest of TutorFlow, and the three steps are priced separately on purpose:
- The plan costs 1 credit. Rejecting an idea should be cheap.
- A build costs 15 credits. This is the step that writes and checks a whole working page.
- A revision costs 15 credits. It rewrites the page with your change applied.
That is less than a TutorFlow game, which has to write scoring, levels and a fail state on top of a model. Every content type's credit cost is listed on the pricing page.
Where a Simulation Is the Wrong Tool
Simulations are narrower than they look, and the narrowness is the point.
- There is no score, no winning and no losing. A learner changes values and reads what happens. If you want a leaderboard and a fail state, that is a game, and it is a different button.
- One principle per simulation. A brief carrying three unrelated ideas comes back as a page that teaches none of them properly. Split it and build two.
- Everything runs in the browser with no network access. That is what makes a simulation safe to hand to a class on school hardware. It also means anything needing live data is out of scope.
- The checks confirm the page responds, not that it suits your class. Nothing automated can tell you the difficulty is right for a group of fourteen-year-olds. Open it once yourself before you share the link.
Start With a Sentence
The fastest way to judge this is to use one. Open the mass on a spring, drag the mass slider, and see whether the class you have in mind would argue about what happens next.
Then write your own sentence. Go to TutorFlow Simulations, describe the one thing you want learners to work out for themselves, and read the plan before you build. The plan is where you find out whether this works for your subject, and it costs one credit to find out.
Frequently Asked Questions
What is a TutorFlow simulation?
A self-contained web page that runs a real calculation and redraws itself as a learner changes the values. TutorFlow generates it from a one-sentence description. It writes an editable plan of the model, builds the page from the approved plan, drives every control to confirm the picture responds, then publishes it to a classroom. Learners open it in a browser with no account and no install.
Do I need to know how to code?
No. You describe what you want learners to work out, and TutorFlow writes both the model and the code. It shows you the model as a plan first. If that plan tracks the wrong quantity or ranges a slider badly, you correct it in plain language before anything is built. Changing a finished simulation works the same way.
How is a simulation different from an educational game?
A game has an objective, a score and a way to lose, and a learner plays it to practise something they have already been taught. A simulation has none of those. It is a model a learner perturbs to find out what depends on what. That makes it the tool for the lesson before the rule is stated, not the one after.
