While I’m waiting for Western Digital to replace my external backup drive — it lasted for 9 months, so that’s pretty good, right? — I’ve been finding comfort in using borg (free) (documentation) , wrapped in the LaunchControl UI ($18) to do hourly incremental backups to my flush Transcend SD card.

LaunchControl is still a little too techy for me, but I got it working pretty quickly. Recommended! And the Transcend is fast enough that I my Mac doesn’t hiccup when it’s being backed up to.

Be the first to comment »

There’s a new version of Wordle called Semantle — not one that I “predicted” — that wants you to find the target word by looking not for a chain of spellings but a chain of semantics. For example, if you started with the word “child” you might get to the answer as follows:

1. Child
2. Play
3. game
4. Chess
5. Square
6. Circle
7. Donut
8. Homer

In short, you’re playing word associations except the associations can be very loose. It’s not like Twenty Questions where, once you get down a track (say “animals”), you’re narrowing the scope until there’s only one thing left. In Semantle, the associations can take a sudden turn in any of a thousand directions at any moment.

Which means it’s basically impossible to win.

It is, however, a good introduction to how machine learning “thinks” about words. Or at least one of the ways. Semantle is based on word2vec, which creates text embeddings derived from an analysis of some large — sometimes very very large — set of texts. Text embeddings map the statistical relationships among words based on their proximities in those texts.

In a typical example, word2vec may well figure out that “queen” and “king” are semantically close, which also might well let it figure out that “king” is to “prince” as “queen” is to “princess.”

But there are, of course, many ways that words can be related — different axes of similarity, different dimensions. Those are called “vectors” (as in “word2vec“). When playing Semantle, you’re looking for the vectors in which a word might be embedded. There are many, many of those, some stronger than others. For example, “king” and “queen” share a dimension, but so do “king” and “chess”, “king” and “bed size”, and “king” and “elvis.” Words branch off in many more ways than in Wordle.

For example, in my first game of Semantle, after 45 attempts to find a word that is even a little bit close to the answer, I found that “city” is vaguely related to it. But now I have to guess at the vector “city” and the target share. The target could be “village”, “busy”, “taxi”, “diverse”, “noisy”, “siege”, or a bazillion other words that tend to appear relatively close to “city” but that are related in different ways.

In fact, I did not stumble across the relevant vector. The answer was “newspaper.”

I think Semantle would be more fun if they started you with a word that was at some reasonable distance from the answer, rather than making you guess what a reasonable starting word might be. Otherwise, you can spend a long time — 45 tries to get “city” — just generating random words. But if we knew a starting word was, say, “foot”, we could start thinking of vectors that that word is on: measure, toe, body, shoe, soccer, etc. That might be fun, and would stretch our minds.

As it is, Semantle is a game the unplayability of which teaches us an important lesson.

And now I shall wait to hear from the many people who are actually able to solve Semantles. I hate you all with a white hot and completely unreasonable passion.[1]

[1] I’ve heard from people who are solving it. I no longer hate them.

Be the first to comment »

I upgraded to a Pixel 6 because my Pixel 3’s charging plug no longer worked, the glass on the back was shattered, and battery life was down to 2-3 hours. I decided to splurge on the Pro version primarily because of its superior cameras.

But while I loved the photos the 3 takes, I’m wasn’t as happy with the 6 Pro. I don’t know enough about photography to be able to articulate why I liked the 3’s photo better, other than to say the colors and shadows were richer and deeper. The 6’s photos were clearer and more detailed, but I generally just liked the 3’s better.

Then I uploaded three sets of samples into this blog post and discovered that most of the differences were due to the displays on the two phones. Viewing them on my MacBook Pro, I think I probably like the Pixel 6 at least as much as the Pixel 3….except for the photo of the hand below.

Here are unedited photos from each so you can decide for yourself.

1 Comment »

Our 3 year old was randomly typing on my wife’s Chromebook and somehow made all of the images go wonky. This includes all images displayed in the browser, the system’s wallpaper, and even the icons.

I could not find any mention of this problem anywhere on the Internet, apparently because I insisted on using the word “posterize” to describe the images’ condition. The rest of the world apparently calls this “inverted.” I have been calling inverted images posterized probably since the late 1980s. It has never before steered me wrong. But according to dictionaries and what Google Search has learned from the Internet, I’ve definitely been misusing it.

But first, the solution to the Chromebook problem. I learned this from Iain Tait (@iaintait) who responded to my tweet asking for help. He pointed to this article in Chrome Unboxed. Our granddaughter unwittingly put the Chromebook into “high contrast mode.” Clicking Ctrl+Search+H will undo the little devil’s mischief.

Now, back to how I went wrong.

Posterization apparently was coined in the 1950s to refer to the process of turning a color image into the sort of stylized image often used in posters. Gradations in color are flattened, colors are brightened, and so forth, until the image would have been acceptable to The Beatles in their late psychedelic phase. Inversion is a 1:1 clipping of colors so that the original looks like what I think a color negative of it would look like, but I’m probably wrong about that too.

Here’s an example using a photo of our post-Thanksgiving walk (CC-BY-SA-NC by me).

Original:

Posterized:

Inverted:

But the real point of this post is to let Google Search see a few more instances of posterize, posterized, and posterizing in the same sentence as image inversion and Google Chromebooks so that the next fool who confuses posterization and image inversion when faced with an image inverted by Chromebook will find at least one damn entry that clarifies a mistake that apparently no one else has ever made.

Posterize inverted images Chromebook. Posterize inverted images Chromebook. Posterize inverted images Chromebook. Posterize inverted images Chromebook.

(The transformations are by Pixelmator Pro.)

* * *

Isian has written a lovely post about how our paths crossed after many years.

Be the first to comment »

[SPOILER: Nope. Not the way to do it. Except for the sentence in green.]

My Windows PC has died rather dead. It does not recognize my boot drive, nor does it boot from my two back-up external drives or from the boot USB I prepared a year ago. Thus continues my 30 year streak of never having made a Windows backup — on any at least four different types of media –that I could successfully back up from.

I thus have to make a new USB boot drive but from my Mac because it turns out that I don’t know anyone within driving distance who uses a PC. And, in truth, I only use mine for games.

There are bunches of pages that tell you the Mac terminal commands that should teach you how to load the Windows installer from Windows , but one of the best I found is by Quincy Larson at FreecodeCamp. Be warned: the process is sloooowwww.

My PC booted from the USB stick, after telling the PC, via the BIOS, that that’s the drive to boot from.

The next step is two spend 2.5 days repeatedly trying to get Windows to install onto a hard drive. My directions are: try everything randomly, learn how to use “diskpart” so you can use trial and error to come up with the right partition and formatting for the receiving drive, memorize by heart and then by muscle memory the keyboard shortcuts for walking through the preliminaries of the Windows installation process, convert the receiving drive from MSR to GPT and back again as often as you can all so that at last …

… you can repeatedly be told that there are no partitions on the receiving drive, and Windows has no way of creating or formatting partitions on the disk that you know that you just partitioned and formatted, even if you politely press the “Delete” button first.

Then, when you’re at you’re breaking point looking at the Windows error message that tells you to read system logs that you can’t find and wouldn’t understand if you could, go back to DISKPART, clean (erase) your disk once more, and either create a partition but do not format it, or don’t even create a partition. At this point I honestly can’t remember. Install Windows onto that disk and you’re good to go…

…where “good to go” means to start reinstalling every bloody program because Windows’ Registry is a jealous god.

Thus does Windows continue it’s long of tradition of being great so long as everything goes well, and being a freaking nightmare when anything goes wrong.

Just ask Clippy.

3 Comments »

I move my laptop from my home office to other rooms of the house after I realize, as I do at least twice a day, that my office looks like a hoarder’s storage locker. Rather than cleaning it up, I decided to make it slightly easier for me to make those moves.

The problem is that if I move my MacBookPro (2019) out of bluetooth range of my Apple MagicMouse, the trackpad doesn’t accept input. And that means I have to trudge the entire 15 feet back to my office to get my mouse. That obviously is simply not acceptable, so I instead spent several hours writing a shell script that I can run in a terminal that will toggle my mouse (or any other bluetooth device) on or off.

It requires you to install blueutil, an open source set of command line tools for managing your bluetooth connections. (Thank you, Ivan Kuchin). To install blueutil, in a terminal type “brew install blueutil”. If you don’t have Homebrew installed, you’ll get an error message, in which case install Homebrew. It’s very useful.

You also will need to know the bluetooth ID of the device you want to control. You can find that by going to your Mac’s Settings and clicking on the bluetooth icon, or by running blueutil in a terminal with the parameter “–paired.”

To use this script, copy and paste it into a text editor and save it with a “.sh” extension. To let your Mac know that this is a text file that should be run, not just read, you need to go to a terminal and change its permissions to 755. (For instructions on how to do that and how to run it from a terminal, check here.) (I have bound the crl-alt-com B key to running it, thanks to KeyboardMaestro.)

Note that I am a bad writer of shell scripts. I had to look up just about everything except how to write a comment. Have some compassion, people! And if not that, how about some pity?

#!/bin/zsh
# Toggles a device's bluetooth activity on and off.
# Useful for when you move out of range of your bluetooth
#   mouse and your Mac's trackpad becomes 
#   unresponsive. For example.
# Requires installing Blueutil:
# https://github.com/toy/blueutil/blob/master/README.md
#
# I post this as Creative Commons Zero, i.e., public domain. 
# David Weinberger, Sept. 24, 2020 [email protected]

# Add your device's bluetooth ID here.
# If you don't know it, run "blueutil --paired"
# in a terminal window to find out.
ID="04-4b-ed-d2-ef-23" # Just an example

echo "Toggles bluetooth device. Requires blueutils be installed.";
echo "USAGE: bluetoggle. No parameters";
echo "Device ID: ${ID}"
echo "--------";

# is the device connected?
# Run blueutil and returns the text.
STATUS=`blueutil --paired`

# is the ID mentioned in the status?
# If not, then it hasn't been connected, so there's
# nothing for this  script to do
if [[ $STATUS != *"${ID}"* ]]
then
    echo "Device not found. Edit bluetoggle.sh to add device's bluetooth ID. Exiting."
    exit
fi

# Look for the ID followed by ", not connected"
if [[ $STATUS == *"${ID}, not"* ]]
then
    echo "${ID} is not connected. Connecting..."
    CL=`blueutil --connect ${ID}` # Create command string
    NEWSTATUS=`${CL}` # Run the command
else
    echo "Disconnecting device..."
    CL=`blueutil --disconnect ${ID}`
    NEWSTATUS=`${CL}`
fi

exit

Be the first to comment »

Just over twenty years ago, computer scientists were racing the clock to fix a possibly devastating error brought about by an over-estimation of the pace at which tech becomes obsolete, which is an over-estimation of the pace of change itself. It turns out that one of the two popular solutions to the problem made the same mistake. And now we’re paying for it, but mainly through some annoyances, not the sort of world-stopping calamity that the prior error threatened.

The problem twenty years ago was that software developers had with some frequency thought that storing the year could be done with two digits, so that 1970 would be saved as 70. After all, the program wouldn’t still be used in 2000! Would we also still be driving around on earth-bound cars, or giving poodles ridiculous haircuts? Ridiculous!

But, if those apps were in fact still be used as the new millennium began, then the two digits internally representing the year would be taken internally as 00, which would be likely to confuse a computer that would assume – based on the way numbers work – that 00 (2000) comes before 70 (1970). And 2001 would look like 1901, etc.

One approach developers took to preempt the Y2K (year two thousand) bug was to change the way the programs expressed date data, allotting four digits to the year. We shall call this “the right way.” But it’s more complex than it seems. For example, you may have to find every place in a complex, integrated set of programs where the date is referred to. You may have to recompile ancient code, unearthing compilers from ancient crypts guarded by three-headed dogs. It was a freaking nightmare for many organizations.

The second approach was to write a little code that looked for year dates between 00 and 20, and write an except that takes them as referring to 2000-2020. Most applications aren’t dealing with dates going back to the beginning of the 20th century, so that worked. Chris Stokel-Walter (twitter: @stokel) in his excellent, brief explainer in New Scientist, says that an estimated 80% of Y2K solutions took this approach, known as “windowing”, but which we shall refer to it as the “Please don’t do this” approach.

Well, now it’s 2020 and some indeterminate number of windowed apps haven’t updated the fix. Thus, some traffic meters have stopped working. As Chris writes, “The theory was that these windowed systems would be outmoded by the time 2020 arrived…”

So, exactly the same over-estimation of the pace of tech obsolescence has led to exactly the same problem. Surprise?

It’s not at all clear, however, who has made this mistake. The developers implementing the windowing patch were staving off an imminent, plausible crashing of globally crucial systems. Windowing was a reasonable approach to forestalling this crisis … but only if there was a system – a human system – to remember to allocate resources for fixing the problem that the patch postponed.

Conclusion: “Human system” is an oxymoron.

Be the first to comment »

Note that this is not late-breaking news. It’s breaking news brought to you late: Android contains a really great Google transcription tool.

Live Transcribe transcribes spoken text in real time. So far, it seems pretty awesome at it. And its machine learning model is loaded on your device, so it works even when you’re offline — convenient and potentially less intrusive privacy-wise. (Only potentially, because Google could upload your text when you connect if it wanted to.)

You can download Live Transcribe from the Play Store, but if you’re like me, it will only give you an option to uninstall it. Oddly, it doesn’t show up in my App drawer. You have to go to your phone’s Settings > Accessibility screen and scroll all the way down to find the Live Transcribe option.

Once you turn it on, you’ll get an icon all the way at the bottom of your screen, to the right of the Home button. Weird that it’s given that much status, but there it is.

I expect I will be using this tool with surprising frequency … although if I expect it, it won’t be surprising.

Be the first to comment »

The first in a series of six posts about my experiences learning how to train a machine learning system has just been posted here. There’s no code and no math in it. Instead it focuses on the tasks and choices involved in building one of these applications. How do you figure out what sort of data to provide? How do you get that data into the system? How can you tell when the system has been trained? What types of controls do the developers have over the outcomes? What sort of ways can I go wrong? (Given that the title of the series is “The Adventures of a TensorFlow.js n00b” the answer to that last question is: Every way.)

I was guided through this project by Yannick Assogba, a developer in the machine learning research group — People + AI Research –I’m embedded in at Google as a writer in residence. Yannick is natural born teacher, and is preternaturally patient.

The series is quite frank. I make every stupid mistake possible. And for your Schadenfreude, five more posts in this series are on their way…

.

Be the first to comment »

At long last I’m getting around to updating Joho’s look, for several reasons:

First, my last redesign made the page look like it was designed in 1999, although I actually did the redesign about seven years ago, I think.

Second, CSS has made it easy enough to avoid using a table as the container for all of the contents of the page. That was always a hack but, it was easier than dealing with floats, etc., for a multicolumn design.

Third, it was especially easy to move away from using a table because I’ve given up on a multicolumn design. That has meant for now stripping out some early 2000s goodness, including my link-rotted blogroll, categories list, and archives. I plan on adding back my blogroll, if only as an historic artefact, via an overlay.

Fourth, the page page needs to be more readable in the mobile camera phones that the kids all use what with their Facebooking, and Chapnapping, and all the rest. I have by no means done all the work required to make this page truly responsive (Everyday Chaos is my best attempt at employing a modified a responsive template), but I’m hoping it will be responsive enough for the kids who are used to their Facebroking and Snoopchitting and all the rest.

Fifth, Medium.com‘s elegant design redefined what a blog should look like. Yes, that was like 18 years ago, but I was too busy with my Facebokchoying and Schnapssbapping to get around to it. Medium not only transformed our expectations about the tools for writing a post, but also about the reading experience: easy to take in, elegant, content-focused. So, I have increased the font size, removed the distracting sidebar, and I hope made it a more inviting page.

I’m obviously not done. For example, I haven’t yet gotten around to updating the page that shows a single post. But I’m happy so far with the redesign. Or, to put that more accurately, I’m somewhat less embarrassed by the page.

Until design tastes change again.

Aaaand, they just did.

Be the first to comment »