theodoregray
Two reasons: (1) solder largely doesn't contain lead anymore. It's been completely banned from plumbing solder in most countries, and its use in electrical solder is rapidly declining. (2) As you say, I valued different things than you apparently would. Solder in that position would not connect meaningfully with other heat-related tools, which otherwise form a nice column. I did that for copper and copper alone because there are such a large and attractive variety of brass and bronze tools it was just too good of a category to pass up, and because having a skipped beat in the cutting tools seemed like a good idea. (Four sets of ten cutting tools in a row could get monotonous in the book, so each row is broken up a bit, for flow reasons).
> Also I am pretty sure even-sized wrenches (hex in particular) is a reference to 6-member carbon rings, very appropriate.
Oh My God. Believe it or not, the fact that even-sided wrenches are in the carbon space, and that the image prominently features the one wrench I have that looks most like a benzene ring, is completely accidental. I absolutely did not plan it that way. Thank you for pointing out how brilliant even my subconscious mind must be.
> You could fill halogens exclusively with welding processes.
It comes down to page counts. Choosing 118 categories isn't just to arbitrarily make it fit the periodic table, but also to very non-arbitrarily make it fit the page count my publisher aims for in all of my books (this being #6 in the same visual style). Sure, there could have been a few more or a few less, but 118 plus introduction/conclusion is a good fit, and so why not make it exactly fit the table format? Splitting up welding would have required dropping other things, and on balance I made the decision that all welding has to fit on one page.
> I don't think about it in terms of shape, count, or location even.
Here is where we get into fun vs. deep meaning. The iconic shape of the real periodic table is one of the most recognizable shapes in the world. It's like the Nike logo, the atom symbol, an apple with a bite out of it, etc. It's a shape billions of people recognize. When you take that specific shape, and none other, and repurpose it into a classification of something else, it creates the momentary dissonance that is the essence of humor. Periodic Table of Tools is of course not nearly as funny as the classic Periodic Table of Vegetables, but it's way funnier than the Well Organized and Sensible Classification Table of Tools. Fun sells.
I appreciate your appreciation of how hard this is. It is not until you start actually trying to create a real, specific arrangement of actual tools that you see all the subtle issues. Believe me, if you came up with your own periodic table of tools, you too would end up with compromises and inconsistencies that would get you sniped at on hackernews.
Some day I might create some new rows, because there are a lot of important categories that are completely absent.
What really surprised me, and speaks to the lack of a periodic table tooling configuration community, is that the domain periodictableoftools.com was unclaimed even late last year, well over 20 years into the era when someone surely would have thought of wanting it.
Concrete tools are missing because I just don't do a lot of concrete work. Had I finished the book about a year later I might have put some in, because earlier this year I built a new studio that involved pouring a 3600sq ft slab with the help of my concrete foreman friend and his buddies. So messy! I don't like to write about categories of tools I have little or no experience with, so other things crowded out concrete and masonry tools, other than carbide drills and diamond saws (which are fascinating because of the steel/stone hardness ratio that determines which model blade you want, as described in the book).
(Also, I would name gauge blocks over granite flats as fundamental to precision, but in any case, all measuring tools are in the same row because they are related, just as are the lanthanides and actinides.) They are at the bottom because that's where they fit most naturally. My logic they should be the noble gasses, because they don't change anything. But there were too many I wanted to include. It was anodizing having to move them to a larger space.
The criticism of saws I reject: I split them by material in columns (wood- v.s metal-cutting), and by size vertically (getting heavier/more powerful as you go down a column). Bow saws are under hacksaws kind of out of desperation, but there are at least as many metal-cutting bow saws as wood-cutting. In fact given the popularity of hacksaws, perhaps in modern times that is the more common application of this style of stretched blade.
You have no idea how long I agonized over how to make this arrangement not be arbitrary, while at the same time dealing with some of the realities of how many tools of different kinds there are in different broad groupings.
For example, as you say, wouldn't it make more sense for measuring tools to be the noble gases, since they don’t change anything, unlike basically all other tools? That’s one of the first things I decided to do, because it’s so obvious. But there are simply too many measuring tools I wanted to include: they had to go in the lanthanides and/or actinides, because those are bigger categories. Having a scale for additive vs. subtractive along an axis sounds great, but there are so many more subtractive tools than additive it would never work. Plus concrete tools, for example, didn’t even make the cut. They are literally not in there because, well, I just had too many other things to fit in that spark more joy for me (which is the ultimately the criterion for what I put in).
In general, what I tried to preserve was (a) similar basic function within columns, (b) tools get bigger/heavier as you go down a column, (c) transition metals are all related despite being spread over 10 columns, (d) lanthanides/actinides being similar within rows instead of within columns (because it’s like they are actually all supposed to go in the third column, but that would just make the thing too wide), and (e) there’s a diagonal line between metals and non-metals.
Beyond that I found a couple of small ways I could put in some analogies with actual chemical properties. For example, the halogens, which are hot, fiery elements, is where I put tools that use heat (soldering, welding, casting, 3D printing). But then I ran out of heat-related tools that made the cut to be categories, so I used the rest of that column for some categories that didn’t fit anywhere else: optical tools and toy tools. So sue me. Alkali earth metals (column 2) are not entirely dissimilar to alkali metals (column 1), so I put hammers in column 1 and things you might typically hit with a hammer in column 2. Imperfect, but there you go, and the counts worked out.
The only individual element I could pay homage to was copper, element 29: that’s where I put all the brass and bronze non-sparking tools (which means they are otherwise in the wrong place, because all the other transition metals are cutting tools, but in space 29 I’ve got brass hammers and bronze wrenches. By all means rag on me for that choice too.
I wanted to split single-edge cutting tools (knives, chisels, etc) and double-edge/edge-and-anvil cutting tools (nippers, shears, etc) into the early and late transition metals, but the counts just didn’t work out, so I tearfully gave up on that.
I’m happy someone at least noticed enough to complain, and actually it’s nice that I’m getting pushback on something as erudite as the suitability of my element/tool analogies, as apposed to, for example, my risible opinions on titanium hammers (which are reflected only in the book, not the poster).
As you say, I’m Theodore Gray and I should know better, right? Which is exactly why I just wrote an essay on why I’m right and this is not just an arbitrary jumble of tools! I spent months on this arrangement! OK, weeks, but it was a lot of time. I even wrote a Mathematica program that helped me pick 1- and 2-letter “element symbols” without ending up with any duplicates, even though that resulted in some weird and difficult-to-explain symbols (much as with actual atomic symbols).
In conclusion, please keep the hate flowing and buy my poster and book.
Thank you for coming to my TED talk, Theodore
Why do I, who would have been the one to implement such a feature, not even remember whether it exists or not? Because if it does, it's horribly useless. It's just rarely what you actually want, and if you do want to do that, you can simply select the cell group, or command-A select the whole notebook, and use shift-return to evaluate everything. That's something I do regularly, but only when I need to, and it would be horrible if it tried to do that every time. I think it was Maple that tried such an interface. (Except their's was 2-dimensional: you could move equations/definitions around on the screen, and anything below or to the right would use definitions from above or to the left of it. This interface was one of the things that convinced me that this was a bad idea.)
The situation with GPTs is different: they do not have the equivalent of kernel state. It took me a while to believe it's actually true, but they have infinite amnesia: every time you send an API call, or submit a new prompt in their web interface, the model is starting completely from scratch with no state or memory whatsoever of anything that's happened before.
That fact that it looks like you're having an interactive conversation in which it knowns what you said before is an illusion created by the fact that the entire conversation all the way back to the start is re-submitted with every new prompt.
Given that reality, and the way people seem to want to interact with these things, the use of a physical context rather than a temporal one seemed most natural. We did talk about implementing our own temporal context mechanism to artificially make the LLM appear to have time-based internal state, like a kernel, but I personally hated the idea, and we never implemented it.
Now, it's possible that at some point these systems will in fact start having internal state (i.e. you have your own personal LLM running on a wad of gum stuck behind your ear, and as you interact with it it constantly fine tunes its weights, or something). At that point we might regret the decision, but I feel like this whole world is so unstable and rapidly evolving that it made sense for us to do what's right now, and worry about the future, if there is one, during that future.
Theodore
Jupyter notebooks were inspired by Mathematica notebooks: that's widely known and has been explicitly said in blog posts and elsewhere by the creators of Jupyter. It's no secret… And as the designer and developer of Mathematica notebooks I am flattered and honored that my ideas are now so widely used, even if it is in a product other than the one I wrote myself.
Time for a trip down memory lane. There were several notebook-adjacent products around at that time (1987) but none of them implemented notebooks as we know them today.
At the time I was using Xcode, which had a terminal-like interface that was a plain text document in which you could put all the shell commands you needed to build your app. This could include calls to the compiler, linker, or just any generic shell commands. To evaluate a given command (which could be one or more lines long) you had to manually select the exact range of text you wanted to run, which was kind of irritating. Any output would be placed directly below the input line. Because the document was pure plain text, it had no way of knowing what was input and what was output, or distinguish new output from old. So all the outputs accumulated, and you had to manually select and delete them periodically. It was an improvement over a glass teletype terminal interface, but I thought I could do better for Mathematica (code name Omega at the time).
The solution I came up with was to create a text document that was annotated with “zones” delimited by vertical bars along the right side of the screen (where cell brackets are now). Initially I used different black and white texture patterns on 4 or 5 pixel wide vertical bars. One texture for input and another texture for output.
This immediately gave two huge advantages: first, you didn’t have to manually select the range of a multi-line input. Just putting the cursor anywhere within an input zone and evaluating would evaluate the whole input. Second, because the system was keeping track of what was output, old output could be deleted automatically and replaced with new output.
Initially I called these things zones, but Steve Jobs suggested I use large close-square-bracket symbols instead of patterned bars, and think of them more like cells in a spreadsheet than like zones in a text document. Once you have brackets instead of bars, it’s immediately obvious that you could have larger brackets that encompass two or more cells. So I created cell groups, initially just to group input/output pairs, but of course it’s not a big leap to also have sections and subsections, then different cell types, styles, graphics cells, etc.
These key defining features of notebooks did not exist in any other system at the time, and to the best of my knowledge were not independently developed by other people. All the current notebook systems that share these broad features, definitely including Jupyter notebooks, derive from the original Mathematica notebooks I developed in 1987-8. (There are other styles of user interface, including some used by Maple and Matlab around the same time, that are called notebooks, but they did not share the key features I’ve described. They were and are fine products that I’m sure have both advantages and disadvantages compared to our Notebooks, but unlike our Notebooks, they are not the precursors of Jupyter.)
So can Stephen Wolfram personally claim to have invented notebooks? No, and he doesn’t: he claims that Wolfram Research did, which is true, since I was one of the co-founders of that company, and I invented that form of notebook.
Theodore
Fun fact: when we made the app the song was of course not yet in the public domain, so I made a deal with Lehrer to trade him an iPad for the use of the song in my app. He was already talking at the time about eventually putting everything in the public domain. His main concern was to try to do it in a way that would prevent record companies from ever making money from his work.
Theodore Gray https://theodoregray.com