People and Blogs

Through his weekly People and Blogs newsletter, Manuel Moreale is emerging as a sort of “Aslan of the small web.”

With Phil Gyford, whose ooh.directory is an old-style-Yahoo! directory of “blogs about every topic,” Kagi’s Small Web search, and Marginalia’s search, the personal, non-commercial, passionate part of the web—what you might call “the old web,” or “the web before capitalism blotted out the sun”—is re-emerging from under the bushel basket that Google’s search algorithm has imposed on it in recent years.

(Here’s a fun way to spend an afternoon: enter a word, any old word, into a Kagi search, and flip the “Small Web” toggle. I guarantee you will find interesting things: here’s a recipe for Brussels Sprout Fried Rice and here’s a post about walking through Bulgaria, for example.)

Which is all a circuitous way of announcing that I’m proud to be included in People and Blogs myself this week. I enjoyed the process of answering Manuel’s questions about my blog and my blogging, and the chance to remind myself why and how much I have loved—do love—this space.

Workout with me: a new group class at Kinetic starts soon!

I’ve been working out, twice a week, one-on-one, with Coach Cayla, for the last 22 weeks. That’s almost 50 workouts; a lot of Bulgarian Box Jump Reverse Crunches.

As Lisa will attest, in the early days I used to drag myself home from the gym, exhausted. 

I just wanted to lie down.

This caused a frisson of tension in our relationship.

I’m happy to report that, on leaving today’s workout, I feel energized, spent-but-not-exhausted, and eager to return on Tuesday.

Starting on March 7, 2024, I’m switching from private training to semi-private, as Cayla is launching a semi-private class on Tuesdays and Thursdays at 9:00 a.m.

Maybe you’d like to join, to help make up a group of six?

Cayla assures me that the class can accommodate anyone, from seasoned veteran to I’ve-never-been-in-a-gym. I am proud to vouch for her: she is patient, an excellent explainer, and she knows her stuff.

Lisa and I have worked out together half a dozen times in the last few months, and I can attest that, as much as I benefited from the one-on-one work with Cayla to get me over the hump of schlumpitude, working out with other people brings a whole new element, a hearty mix of competition and solidarity. I’m looking forward to it.

If you’re interested, or want more details, you can contact Cayla through her website

It will be fun!

Lockup

In letterpress, lockup is the process of assembling  a jigsaw puzzle of rectangles around the type to lock it all in place, with the ultimate goals of (a) having it not fall out while on the press and (b) having all the type level, so that it gets inked evenly and prints uniformly. 

The pieces of this jigsaw puzzle, beyond the type, are a mixture of wooden and metal furniture, mixed with properly-sized-to-the-type metal spaces of varying widths. 

In a not-completely-outfitted shop like mine, lockup  can be particularly challenging, due lack of diversity and quantity of furniture, but lockup is always challenging anyway. 

Here’s a photo of my lockup for one side of the February This Box is for Good box. This was, indeed, challenging, as space was at a premium, yet the specific distance between LOVE and FOOD needed to be exact, and BE THE needed to be centred under FOOD. I’m proud that I managed to pull it off, even if it is a little Frankensteinian. 

What is the percentage of squirrels that are online?

The other morning at breakfast we were wondering what percentage of squirrels are online. 

We were wondering this because, the previous day, I’d left a This Box is for Good box filled with chocolate on the doorstep of a friend, and Lisa wondered if, before it got to her, it might be captured by squirrels. And so we wondered if, having so-captured the box, the squirrels would follow up by registering the box on our website, as the box instructs.

This led us, as you can understand, to wonder “What is the percentage of squirrels that are online?”

So I asked Perplexity, the AI-powered search engine that New York Times writer Kevin Roose recently took for a ride by way of seeing whether it might out Google as his go-to place to find things out:

Hundreds of searches later, I can report that even though Perplexity isn’t perfect, it’s very good. And while I’m not ready to break up with Google entirely, I’m now more convinced that A.I.-powered search engines like Perplexity could loosen Google’s grip on the search market, or at least force it to play catch-up.

Here’s how Perplexity answered:

The question seems to be based on a misunderstanding or a joke, as squirrels, being non-human animals, do not have the capability to use the internet or be “online” in the way humans do. Squirrels are small mammals known for their bushy tails and propensity for climbing trees. They are found in many parts of the world and are often seen in urban areas, but they do not have the cognitive ability or physical means to access or use the internet.

I love that reply six ways from Sunday.

You don’t need to be fluent in Spanish to order coffee in a Spanish café

François Chollet, software engineer and AI researcher at Google, writes on Twitter (via Simon Willison):

People seem to be falling for two rather thoughtless extremes:

1. “LLMs are AGI, they work like the human brain, they can reason, etc.”
2. “LLMs are dumb and useless.”

Reality is that LLMs are not AGI — they’re a big curve fit to a very large dataset. They work via memorization and interpolation. But that interpolative curve can be tremendously useful, if you want to automate a known task that’s a match for its training data distribution.

Memorization works, as long as you don’t need to adapt to novelty. You don’t *need* intelligence to achieve usefulness across a set of known, fixed scenarios.

(LLM is Large Language Model, the most well-known gateway to which is ChatGPT; AGI is Artificial General Intelligence, the notion of a computer that can think like people can).

Outside of the realm of AI, his statement “You don’t *need* intelligence to achieve usefulness across a set of known, fixed scenarios.” echos what I tell people who are skittish about travelling to a destination where they don’t speak the language (and where their language isn’t commonly used): you don’t need to be fluent in Spanish to order coffee in a Spanish café, you can just say “café, por favour.”

Similarly, you can point at a baguette in a Parisian bakery, and say “un, s’il vous plaît.”

You can do the same thing when you’re pointing at a flavour of ice cream.

You can accomplish a surprising amount by memorizing a few key phrases and using them in known, fixed scenarios. We once had a lovely bus ride from Venice to Ljubljana, on the Florence-to-Sofia run, with a very kind group of Bulgarian workers returning home. We shared food with them, and they with us. 

As long as you don’t need to adapt to novelty,” which is a lot of travel (a bakery is a bakery is a bakery), everything goes fine. Doesn’t help when the bus stops in the middle of nowhere and everyone’s told to get off, or when the bakery is out of bread, or you need a membership to shop at the Italian coop grocery. Or when your Spanish waiter asks you what type of coffee you want. But, most of the time, in most places, you can get by just fine using the dataset you carry around about the basics of how the world works.

Why can’t I produce a report from a spreadsheet?

I’ve been scheduling interviews for the last few days, and part of that job has involved taking a spreadsheet exported from Typeform (where we collected responses to a questionnaire from candidates) and trying to find a way to make a report from it.

You’d think this would be easy, the kind of thing people do in offices all day long.

All I want to do is take this:

First Name,Last Name,City,Start Date
Peter,Rukavina,Charlottetown,2024-04-01
Sully,Sillerson,New York,2024-05-01

And turn it into this:

First Name:  Peter
Last Name:   Rukavina
City:        Charlottetown
Start Date:  2024-04-01

With one person’s details on each page.

I can sort of do that by loading the spreadsheet into Numbers and selecting Table > Transpose Rows and Columns. This gets me halfway there; from that point, I can selectively hide and show columns, print each, and get where I want to go.

Screen shot of Numbers, showing four columns and two rows, with one person's responses to a form in each row.

A screen shot of the same spreadsheet, but with rows and columns transposed.

Screen shot of the same spreadsheet, but hiding one column, so we're just seeing one person's response.

But surely.

My second real computer job, when I was 16 years old, was at the Canadian Tire store in Burlington, Ontario. I started off working there on the sales floor, selling Commodore VIC-20s to the unsuspecting; eventually I was moved upstairs to the comptroller’s office, where a copy of the newly-released Lotus 1-2-3 had recently arrived. 

Lotus 1-2-3 was amazing. It’s hard to convey just how amazing it was to my teenaged eyes, but I loved it, and loved figuring out how to make it do all sorts of amazing things. Canadian Tire was able to learn much about itself because of my work, from how productive their mechanics were to gaining insights into wages vs. sales. Along the way I learned a lot about Canadian Tire, and business, about spreadsheets, and about how to be a computer consultant.

Using Numbers, more than 40 years later, isn’t much different than using Lotus 1-2-3 back then. There certainly hasn’t been 40 years worth of innovation or creativity applied to the category.

The underlying issue here, of course, is that Numbers is a spreadsheet, whereas what I really want is a database manager. Indeed, I made a good living from being a dBASE manager for much of my 20s. 

Producing reports was one of the things dBASE did well, and what I really want is its modern equivalent.

But apparently there isn’t one. Whereas VisiCalc begat Lotus 1-2-3 begat Numbers, it seems like FileMaker and dBASE and their ilk are a category of consumer software that died on the vine.

(Technically FileMaker didn’t die, it just became something else: an $800 piece of enterprise software that can, indeed, produce a report from an imported spreadsheet.)

What do people do these days when they want to keep track of generic things, search for those things, report on those things, share those things?

When did computers on our desks stop being able to do that?

Pages