while True: learn() (PC) Preview

There will be plenty of comparisons with Human Resource Machine; comparisons that are very important to understand where while True: learn() succeeds with its concept, and where it - sort of - fails with it. Having said that, though, the central idea is exactly the same: there is a bunch of data on the right (colours, letters, or shapes), and through the use of a visual interface that simulates algorithm creation, you will send this data to the appropriate "box" - red letters to the red one, blue to the blue, and so on, and so forth.

Now, while those completely unaware of what programming is will be somewhat baffled once the game begins, the interface is actually easier to use than the one in Human Resource Machine, as it's a simple matter of dragging-and-dropping pieces of code onto a grid, and then connecting them. These are called nodes, with the "beginner" one dividing the incoming data between two outputs. Therefore, if wanting to, say, discard all red bits, you create a node where one line sends those to the bin, while everything else is sent to the final output.

More node types are added as one gets deeper into this "adventure," and, as expected, the tasks get progressively harder. This, however, is the perfect time to mention a flaw, and a pretty big one at that. The UI is a piece of cake to use, but it's not so easy to understand. No, it's not badly made or anything, but, unless into programming, most will stare at the tutorial screens, scratching their head for quite a while. In comparison, Human Resource Machine had some fierce puzzles to solve, but it was always easy to understand what was on the screen.

The way this handles success is also another problem. More specifically, while True: learn() rewards for the number of items that reach the output, the accuracy (the right data in the right box), time spent for the programme to conclude, and the money spent on servers, yet this needless complexity is more realistic than entertaining. By contrast, Human Resource Machine condenses all these goals into one: the amount of steps the programme will go through before completion, with less steps being the better choice.

This leads to the title's main issue: it feels less like a game, and more like an educational piece of software that wants to introduce people to machine learning - not just because of the few snippets of history this offers every now and then, but mainly because of the overall look. Besides the intro screen, where the protagonist is seen right in front of his monitor, with his feline buddy sleeping on it, everything else looks like how a "real" programme would look, when this would really benefit from a more... videogame-style art approach.

In the end, and, again compared to Human Resource Machine, this lacks the necessary character to make, a) those familiar with programming prefer it over actual coding; b) those who don't know anything about the subject have some good 'ol puzzle-solving fun. The good news? The development team's love for its creation is rivalled only by its respect towards the community, as it is very active in Steam's forums, taking notes, and promising fixes while at it. Hopefully, the final release will reflect that.