to teach my course if it’s not mandatory

The conversation today began with a technical question: should students complete the initial modules on GitHub, or can they choose Tilda, Cargo, Figma, or any other tool that works for them and won’t fall apart during review? It quickly became apparent that the question about the platform masked a larger issue concerning the course structure itself: what does the instructor, who inherits someone else’s lectures, assignments, and several hundred slides—some of which contain valuable insights, and some that simply create atmosphere—actually receive?

The course spans an entire academic year and is built around four projects: an interactive web publication, a landing page for a conceptual artifact, an e-commerce store, and a mobile interactive system. These are framed by lectures, short in-class exercises, homework, and detailed consultations; integrating all of this into one cohesive unit is nearly impossible, even if one were determined to act like a proper instructional designer.

I could have said that the course contains a mandatory core around which each participant is free to develop their own methodology. That would have sounded professional, reassured my younger colleagues, and immediately proved untrue. There is nothing immutable here: the course was structured around my way of thinking and presenting, so I need transitions from Heidegger to Figma, from Deleuze to web form, from Harman to commodity—it is through these conceptual lenses that I approach a project. Another instructor might find these challenging, requiring excessive preparation time, or simply clashing with their temperament, and then they can be removed. Heidegger is not a license key required for Auto Layout to function.

One can take an individual slide, an exercise, a diagram, a criterion, or a line of reasoning; one can change the order of topics, assemble their own presentation, shorten the theoretical components, and add the theory you actually know how to apply. Even the sequence of sessions bears the imprint of my habits; I found no hidden pedagogical law within it. This text also lacks any administrative authority. This is a description of my current position—one that will very likely change after the next semester, a few more reviews, and another Figma update that will once again hide a familiar button in a place the developer won’t intuitively find.

However, having the option to deconstruct a course doesn’t automatically make any build successful. It simply shifts the responsibility. When an instructor removes my philosophical lenses, they must decide through what lens they themselves will teach students to see, differentiate, and justify. If, after deconstruction, only a sequence of operations within the program’s interface remains, the course quietly dries up into a tutorial video—and in most cases, the video will explain the operation more quickly, patiently, and allow the user to review confusing sections at 0,75x speed.

Student complaints are well-known: «They didn’t teach us Auto Layout in Figma, but they did overload us with Deleuze and Heidegger.» There is an uncomfortable measure of truth in this criticism. One cannot label a course as conceptual and then use that to excuse a technically incompetent mockup. If elements break apart, text spills out of containers, and the same control exists twenty different ways across twenty screens—conceptual depth won’t glue it back together.This, however, does not mean that a significant portion of class time should be spent giving tours of the current Figma interface. The tool evolves faster than the curriculum, and the skill of finding solutions is already part of professional practice. It is permissible and beneficial for a student to independently figure out how to build a component, configure adaptive behavior, or publish a website; the instructor sets the required standard beforehand, shows how to verify it, and intervenes only when a gap reveals a misunderstanding of the underlying principle.This is particularly evident with components. In a small conceptual project, it isn’t necessary to immediately turn every utterance into a component—religious mandates would just add needless busywork. However, a repeating element that is destined for changes is best organized systematically from the start. Otherwise, a week later, the student discovers twenty orphaned copies and manually corrects the same button on every screen. At that point, components cease to be an abstract «best practice» and become personal hygiene, saving several hours of life.This skill is assessed without a separate exam on button placement. It is enough to ask the student to change one repeating pattern, replace long text with short text, or restructure a screen for a different width—and then observe whether the change propagates through the system or if the student fixes every fragment manually. The instructor sets the threshold, diagnoses the cause of the failure, and sometimes opens the file to demonstrate a specific technique. The mastery of the interface itself remains with the student, who can choose the appropriate explanation, pace, and depth.

To ensure this agreement does not appear as a trick, it should be stated at the very beginning: «We are assessing the technical quality of your work and will require you to master the necessary features. Classes are dedicated to project-based solutions because documentation and training videos explain the program’s interface better than lectures, and they are not yet equipped to discuss your concept, scenario, or the implications of your choices.» After such a statement, a complaint such as «we weren’t taught this» becomes a discussion point regarding the distribution of responsibility; it no longer serves as proof of the program’s hidden failure.

The general tool is convenient for the instructor: it’s easier to access submissions, technical requirements are simpler to standardize, and unexpected failures are less frequent. The recommended pathway is also useful and advisable for a beginner group—it alleviates initial anxiety for those who are not yet able to compare five platforms and quickly get lost in a digital swamp featuring three free tiers, two expiring trials, and a domain that might be deactivated tomorrow. Such a pathway can be presented as a reliable guideline.

Adjacent to this path, there must be an exit. A student can assemble a publication on Tilda, Cargo, in raw code, or within an environment the instructor has never used, provided they can justify their choice, demonstrate a working result, and do not require the project consultation to turn into a 24/7 support line for some exotic engine. By the time of graduation, a specific service may change unrecognizable, lose its accessibility, or be superseded by something new; the ability to select the right tool for the job endures far longer than any particular platform.

The freedom of the tool necessitates common criteria for the outcome. The comparison begins with how effectively the publication solves the posed problem, whether it sustains the chosen genre, how technically sound it is, how it manages attention, and whether it retains meaning after encountering a real reader—the virtuosity of using a single panel is secondary here. If a platform limits a solution, that limitation is part of the project narrative: the limitation can be accepted, worked around, or transformed into a deliberate constraint, but it cannot be retrospectively presented as intentional design.

There is also a less polished reality: access is unevenly distributed. One student may have paid for ChatGPT, Claude, Midjourney, and several other services whose names change faster than I can keep track of; another might only have a free account and a bank card that half the services don’t accept. Therefore, a subscription alone cannot constitute an educational advantage worthy of a grade. Instead, we should compare the quality of selection and the way available resources are utilized.

The older version of the course included many large workshops: screencasts, lyric videos, letter portraits, and other exercises that were once useful and quite impressive. Year after year, students became less and less enthusiastic about completing them. The issue is likely not just declining morals. A major exercise requires almost as much attention as an independent project, but it ends in a standalone artifact that leads nowhere further. The student already has the core work, wants to move it forward, but the instructor hands them another beautiful, rudderless textbook.

These kinds of assignments can be archived and retrieved when they are certain to resolve a detected problem. If someone wants to conduct a separate lab on visual poetry—that’s wonderful. Somewhere, a lyric video might prove to be the best way to grasp the rhythm of typography. Imposing a mandatory nature here adds nothing.The short in-class interactives operate differently. They last only a few minutes, require an immediate trace—a diagram, a solution, a signature, a micro-prototype—and allow the theory to immediately become intuitive. Meanwhile, homework continues the main project. This creates a rhythm where the session provides local tools for vision, and the project incorporates their consequences. If an exercise does not alter the student’s subsequent work or reveal a new way of seeing, it should have very strong justification to take up half of the class time.

The first module, based on habit, is called the «longread, ” although many excellent works have long since moved beyond this genre. A longread is meant to facilitate the comfortable reading of a lengthy text, sometimes including multimedia insertions. Students submit interactive zines, web installations, visual poetry, non-linear narratives, and almost content-light spectacles where text yields to movement, rhythm, and state. Calling all of this a longread is only appropriate in the same way that we call a phone a device where the act of making a call occupies one of the most humble functions.A more precise framework is the interactive web publication. It accommodates various genres, but it still demands publication: the work has an audience, an organized duration, a typographic body, interaction rules, and a reason for existing in the browser. The instructor can narrow the scope to authorial text or, conversely, allow a web installation with nearly zero content. The decision depends on what the instructor intends to teach through the project.

If a student uses a language model to generate text, the focus of the evaluation shifts to the editor’s responsibility. Who gathered and verified the material? Who structured the composition? Who spotted stylistic tics? Who can rewrite a paragraph and explain its purpose on the page? Formally establishing an «AI percentage» is hardly helpful here. Mastery of the text is demonstrated in conversation: the author can differentiate its parts, recognize weaknesses, adjust the tone, and cannot simply defend a random phrasing because it looks good in the layout.

In this module, typography is also better taught through the perception and structure of form rather than mandatory memorization of antique classifications. The student needs to see where the movement of writing still lives within a letterform, where the shape has become static, how a typeface accumulates cultural reputation, and why two beautiful designs placed side-by-side might form an unappetizing pair. Mashed potatoes and french fries are perfectly fine on their own; serving one as a side dish to the other is odd. It is similar with typefaces: similarity or contrast must have a plastic and semantic reason. The history of type helps reveal that reason.

The freedom to choose a topic often seems like an unconditional gift until a student selects something intensely interesting, conceives an idea, and then confronts the necessity of expressing it. References, questions about the audience, and the requirement to construct a coherent scenario appear—and the initial excitement quickly cools. It seems as though the mere fact of a good topic should have done the heavy lifting.It’s not enough to generate an idea; you must also communicate it so that it reaches its intended destination. An idea you fail to convey exists nowhere: it doesn’t become deeper because the viewer didn’t understand it, nor does it become more radical due to the absence of an audience. Even a project about contemporary art is made for specific people. You don’t have to explain it to everyone; you just need to decide whom you want to enter into a dialogue with—what these people already know, where they might get lost, and what kind of experience they’ll agree to engage in beyond the first screen.

Therefore, complete thematic freedom requires intermittent collisions with reality. At the start of the project, the student defines the intended audience and the project’s promise; then, quite early on, they present a small but functional fragment. Following this, they verify whether the intended concept is legible without a five-minute authorial preamble. The reference here is not for finding a surface to copy. Instead, it demonstrates how another person’s work has translated a complex theme into a composition, metaphor, navigation, or sequence of actions.

The instructor aids this translation process without completing the project for the student. They pivot the conversation away from the abstract desire for «something more original» toward the precise point where the thought has lost its form: the metaphor conveys nothing, the typography conflicts with the content, the interactive gesture merely adorns the page but does not alter the reading experience. This—driving the concept to a state where it can leave the author’s mind—is one of the primary functions of consultation.

In discussions about student projects, the word «banal» begins to mean far too much. A familiar icon or a standard navigation scheme can be a good solution: the user recognizes the pattern and interacts without needing extra instruction. The same familiar trope, automatically carried over from an initial iteration and unrelated to the task, already looks cliché. An unfamiliar solution sometimes opens up a new scenario, and other times it just forces the user to relearn how to press a button because the designer wanted to immediately differentiate themselves from the norm.

It is useful to distinguish between these cases based on two questions: how familiar the solution is and what justifies it. A familiar solution with a clear reason functions as a pattern or an honest MVP; without a reason, it remains a default. An unusual solution with a clear rationale becomes an invention, while one without it creates deviation for the sake of deviation—an effect that wears off after the first attempt to use the project.

Students especially often start with deviation, before they have even figured out how typical human-interface interaction works. Here, one must temporarily defend the banal, even though the word sounds suspicious. Control elements can remain familiar as long as the originality lies within the value-delivery scenario, the structure of the narrative, the relationship between the system participants, or its visual voice. Novelty does not have to reside inside every button like a tiny imp.

Resource constraints can also justify a simpler solution. If a student built an MVP in a short timeframe, opted out of secondary mechanics, and retained a working scenario, that demonstrates mature pruning. During the defense, you must state the constraint and clarify exactly what was deferred. The phrase «we didn’t have time» explains nothing; the statement, «we were testing a specific hypothesis, so we brought this section to a working state, while consciously leaving the rest outside the prototype’s scope, ” already describes a deliberate design decision.

AI tools can be used in the course, and in most cases, you should at least try them where they fit the task. Banning them quickly turns learning into a mere reconstruction of a fading profession. However, the mere fact of using a model does not prove the project is contemporary; you can assemble a complex pipeline to generate an expensive cliché, which you can then neatly drop into Figma.

Strong automation changes the demand placed on the outcome. If generating variations becomes inexpensive, the sheer quantity produced no longer impresses. The value shifts to curation, problem definition, assembly, review, and refinement. The student should document which tools and versions they used, what they assigned to each, why they assembled the sequence in that specific way, which variations they discarded, what criteria they used to make a final decision, and what they refined manually or with a subsequent model. For imagery, this might be the path from concept art to refinement and integration into a system; for text, it could be the process of research, drafting, editing, and source verification; for interface design, it could be prototype generation, structural analysis, logic correction, and testing.

A prompt can only enter a design system as a reproducible tool; a talisman that once generated a beautiful image has no place there. In such cases, the model and version, the prompt structure, constraints, acceptable variability, and subsequent processing are documented. Verification is straightforward: another team member follows the documentation and receives material generated by the same system; a vaguely related marvel indicates random luck.

This log preserves the trace of decisions. Ten pages detailing how a model didn’t cooperate won’t make the final work better. A «compilation of sad songs» about prompts might prove valuable diagnostic material during consultation, but standard quality criteria still apply when viewing it: does the project solve the problem, is it conceptually and visually cohesive, does the interaction function, is the system robust, and can the author take responsibility for the outcome?