A Tale of Two Kits

This entry is part 5 of 5 in the series Ribbonfarm Lab

I’m working on two kits at once: A Lego Technic Perseverance kit, and a “hello world” spoon using a beginner’s whittling kit. They could not be more different, yet I can see why both deserve to be called kits. Both come complete with all the tools and materials needed, and guidance on what to build. Both are designed with an initial outcome in mind, but can also be used to build lots of other things the designers didn’t think of. Both embody design dispositions towards accessible design spaces. They are weakly opinionated engineering artifacts.

But there the similarities end.

The Lego kit has no tools, complete instructions, over 1000 parts, and a very credible assurance of getting to the promised outcome with ordinary skills. The manual is unambiguous pictures, no text. If you think the instructions are wrong, it’s probably you that’s wrong. The kit is built in 4 modules, and the parts for each are separated into 1-2 bags.

I’m part way through module 1. They’ve wisely front-loaded a small win in module 1: the main subgoal is that you assemble part of the rover chassis, but first you build the complete Ingenuity helicopter. They’ve really shaped the motivation curve. Here’s where I am:

Close-up of the helicopter:

The whittling kit comes with 3 knives, a sharpening/honing strop, some wood blanks, cut-resistant gloves, a pattern for making a bunny (I’m making a spoon instead; seemed simpler), no instructions, and no assurance of getting anywhere. I started to wing it, but struggled, so I had to find some YouTube videos to learn basic techniques. Here’s how it started:

And here’s how it’s going:

I got the rough profile done using basic techniques from YouTube, but am still mostly winging it. I am struggling with the cavity of the spoon. I might need more videos.

The underside of the spoon is coming along better.

One of the reasons I’m doing this experiment of parallel kits with very different dispositions is that I’m interested in the duality relationship between kits and protocols. How do codified behaviors operate on grammatized spaces of stylized toolsets crossed with a finite vocabulary of parts? My theory is that kits induce protocols and vice versa. Or more poetically, kits tell protocols how to codify, protocols tell kits how to modularize. The specifics of that strange loop define the disposition of the kit.

The Lego assembly protocol is so mature and finished it’s hard to do anything irreversibly wrong. Building with Lego is almost isentropic. If you have to force a fit, you’re probably making a mistake, and it takes significant strength to make one. I don’t think a child could even break a part. The protocol is almost foolproof and nearly deterministic. All the skill and knowledge needed has already been absorbed into the parts themselves. This is why you don’t need tools, critical thinking, or imagination; just basic dexterity. (Well, the brick separator cheat tool is sometimes helpful). Lego is elementary training to be a factory worker. Whittling is training to be a crackpot survivalist prepper.

Whittling is of course about 100x older than Lego. I assume people in the Bronze Age idled away time whittling with their crappy bronze knives. But it is far less mature and finished in a sense. Every action is irreversible and all mistakes need creative fixing. At first sight, the activity seems too intuitive and natural to have protocols. Instead of being foolproof it almost seems to require a narrow sort of savant-like enlightenment to do at all, let alone do well. You either have a “feel” for it or you don’t, and if you don’t, structure won’t help much. If you do, you can develop the feel into a skill. I don’t think I have a strong feel for this, but I have some feel. I suspect I’ll patiently and laboriously get to passable mediocrity at best. I won’t get to refined craft, but maybe I can make up for that with some interesting design ideas. The activity seems to be dominated by a very tropey traditional artisan design space. Everybody seems to make the same animals, faces, etc. Surprisingly, despite being so late-industrial, Lego seems to feature more imaginative builds. Perhaps taking the skill out allows more imagination to flourish. Artisanship tends to get in the way of art.

As I found out the hard way, some structure helps. Just trying to Zen your way to Spoon does not lead to great results. You cannot get to Spoon by using carving away Not Spoon. Beginner Mind is not enough. You need Beginner YouTube. YouTube tutorials offer a lot of tips and heuristics, but also a few imperatives I’d call protocols.

Here’s one: you hold and control the blade with your right hand (if right-handed), but push it with your left thumb, with the rest of the left hand holding the piece. This makes for very safe whittling, since it’s much harder to lose control and cut yourself. You don’t have to do this, and sometimes you have to make exceptions for awkward chipping needs, but it’s a good default.

What the YouTube tutorials seem to teach is a few stylized techniques, key safety behaviors, and pointers on avoiding mistakes (fixing mistakes is hard). Beyond that it’s just demonstrations of whittling specific traditional objects (which are kinda ASMRish to watch). This makes sense. Due to the nature of the tools, there is an infinite number of ways to bring blade and wood together, most of them clumsy, ineffective, and unsafe. The “kit” aspect really lies not in the vocabulary of parts and joins, but in learning a few foundational moves and sequences (3-4 in most of the tutorials, all somewhat intuitive; I found myself doing them before looking at the videos, but sloppily, and mixed with less effective hacks — is this where “hacking” comes from? It certainly removes wood but usually in a way that creates technical debt in the wood). It feels closer to the kihon-kata-kumite skill stack of Asian martial arts where Lego feels more like a basic, finite literacy. I could use a few wax-on-wax-off Miyagi whittling katas right now to finish my spoon, but I don’t think I need to learn anything more to finish the rover.

Unlike Lego, which has no tools or real skills needed, whittling requires not just serious tools (one or more knives, sandpaper, and a strop), but maintenance behaviors (sharpening, honing) and judgment to switch between tools, as well as between tool use and tool maintenance (“is the knife sharp enough or am I whittling wrong?”). Right now, I’m following YouTube advice and mindlessly stopping every 15 minutes to use the strop, 20 strokes on each side. A feedforward maintenance protocol that will hopefully become feedback driven as my actual sensitivity to sharpness improves.

It’s interesting that the Lego kit is all parts, but the whittling kit is basically all tools. It comes with some cuboidal bass wood blanks (soft, with straight grain with no knots), but in theory you don’t need those. The poetically ideal way to whittle, approved by Walt Whitman, is with wood you randomly pick up, using only one old knife that you sharpen with a stone found in the wild and some spit. And you only carve designs that honor the contours and knots of the found wood or something.

What’s more: the Lego parts basically never wear out (bricks from the 70s will mate with bricks from today) while the whittling tools wear out steadily and need constant maintenance.

Last point for now, though this tale of two kits will continue: The Lego kit and assembly protocol is very complete and legible in relation to its stated goal of getting to a completed rover model, but the extended whittling kit (including augmentation with YouTube videos) is very incomplete. It codifies only 20% of what you need to know to get 80% of the way to a hello world project like a spoon. The remaining 80% is illegible tacit knowledge your muscle memory has to pick up by hacking anyhow.

On the flip side though — and this is probably a consequence — while building a Lego model, you have no sense of progressing from rough, coarse, strategic actions to refined, detailed actions. It’s all at the lowest level of fine detail, and you feel like a puppet in the hands of an inscrutable kit designer who has already done all the strategic thinking and planning. The model comes together via a series of very efficient but non-hierarchical moves in a way that’s much smarter than your execution awareness. One move might add a bit of cosmetic detail, the next move might construct a load-bearing spine of the design, and the third move might add a dispensable flourish. You only get what’s going on in moments of hindsight as big subassemblies come together.

The consolation of the low level of definition with whittling is a high sense of autonomy and agency. Instead of the designer’s awesome rover model, constructed by you, puppet, you get your own crappy spoon.

I’m not yet sure which I prefer: pawn on someone else’s chessboard, or lord of my own tic-tac-toe board.

Series Navigation<< Lego Soup

Get Ribbonfarm in your inbox

Get new post updates by email

New post updates are sent out once a week

About Venkatesh Rao

Venkat is the founder and editor-in-chief of ribbonfarm. Follow him on Twitter


  1. I’m reading this through a software development lens. Software libraries and frameworks (interestingly, Apple calls them “kits”) have a lot in common with your two kits. Although they exist along a spread on that same spectrum, not quite going as far as either of your examples.

    A low level C library feels more like whittling, and you need to know a lot about the system and your tools to get anything meaningful done. However, it’s not quite as “low-level” and unforgiving as whittling.
    A high-level framework like SwiftUI has everything in the parts (components) and you just need to assemble them, including quick wins to get something on the screen quickly, but it’s not quite as high-level and restrictive as Lego.

    An analogy that comes to my mind is: Lego and high-level programming are convenient because everything aligns to a grid. As long as you are ok with the smallest size of a grid cell, the smallest brick size or “sampling resolution” so to speak, you’ll love creating with that constraint. But you can never go below the threshold without getting the tools out and your hands dirty.
    Whittling and low-level programming are more challenging, because they require tools and dirty hands, but the resolution you work in is only limited by your skill. If you have little skill, that can feel daunting (and makes Lego blocks so much more appealing), but as you acquire more skill, you’ll appreciate your less limited agency even though it’s less convenient and more work, because you are not just building a thing, you are also building yourself.

    I suspect that points to an interesting problem with the trajectory we’re on in software (and the tech industry at large): We clearly favor the Lego approach because of convenience, quick wins, etc., but the fixed sampling resolution (or is it even getting coarser due to rising complexity?) prevents us to get to really satisfying results. And we all feel more like pawns on someone else’s chessboard.
    What we all would like is to have the freedom of the whittling approach, but the burden to acquire enough skill first conflicts with our laziness. Over time, we end up with lots of factory workers with some creativity to want to build new things but not enough skill to do it beyond the components that we already have. Everything snaps to the grid of the world we have built so far, until some rare craftsperson comes along, who meticulously whittled their way beyond the resolution of the grid and impresses us with a piece of art, which, of course, is not scalable. Meanwhile, we’re also recruiting generative AI to help us explore the still infinite possibility space of the grid we live in.

    Also seems to fit into your mediocrity framework, if we assume that coarse resolution Lego blocks lead to average designs that lack the fidelity of a whittled design executed with skill.

  2. Could you whittle a Lego?

Leave a Comment


This site uses Akismet to reduce spam. Learn how your comment data is processed.