Using Science to Decide What Article to Write Next

I will walk you through an example of how you can use this method to decide what Medium article to write next.

Note: I created an Excel template you can reuse for any decisions in the future — see the link at the bottom of the article.

Here are the main things you need:

1. A list of options

This is your list of article topics to choose from — for our example let’s say the choices are:

• Personal goals
• Technology
• Pets
• Travel
• Finance

2. A list of decision factors

These are the different criteria you want to use to make a decision — for writing a new Medium article, these could be things like:

• Research time — how many hours it would take you to research the topic
• Writing time — how many hours it would take you to write and edit the article
• Chance of getting curated — your best guess about your odds of having the article curated by Medium editors
• Expected popularity — on a scale of 1–10, how likely is your article to go viral and make you lots of money

Of course, there can be many other factors that go into your decision of what to publish on Medium — use what makes sense for you.

3. Weight of each decision factor

Next, you need to assign a weight to each decision factor — how much does that particular factor matter compared to the others. The weights should of course add up to 100%.

You also need to specify if a factor will positively or negatively impact the result. For example, the writing time can be a negative factor — the longer it takes to write an article the less likely you are to choose that topic. Others, like the expected popularity, are positive — a higher number means you are more likely to choose that topic.

For our example, let’s use the following:

Putting everything together

Here is the full list of options and the values I assigned to each of the decision factors in the example:

And that’s it! The model now has everything it needs to run a TOPSIS analysis, and based on these parameters it can tell us the winner:

Warning: this section contains math — if you’re not interested, skip to the bottom for a link to the Excel template!

For the nerds like me out there, here’s a step-by-step explanation of how the analysis works:

1. Normalize the decision matrix

This is needed because the values of the decision factors use different dimensions — some are in hours, others in percentages, others in a 1–10 range. The result is a similar matrix with all values normalized:

2. Apply weights to each parameter

Essentially you multiply each value in the normalized matrix with the weight of the corresponding decision factor. The result is a weighted matrix that looks like this:

3. Calculate the best and worst alternative

Next, you calculate the best (V+) and worst (V-) alternative based on the parameters you specified. You do this by taking, from the weighted normalized matrix:

• For V+: the maximum value of the positive factors, and the minimum value of the negative factors
• For V-: the maximum value of the negative factors, and the minimum value of the positive factors

The result is a table like this:

4. Calculate the geometric distance between each option and the best and worst alternative

Imagine each option is plotted as a point on a two-dimensional axis, along with the best and worst alternative. The goal is to calculate the distance between each option and the best and worst alternative, respectively. Here is how that looks in two-dimensions:

The formulas used above do the same calculation, but in n dimensions, where n is the number of decision factors. The result is a table similar to this:

5. Calculate the similarity of each option to the worst alternative and rank them

Each option is assigned a score that indicates the similarity of that option to the worst alternative — the best option is the one that has the least similarity to the worst alternative — it is farthest from the worst and closest to the best.

The score and the final ranking of each option is shown below — Travel wins.