What I learned last week

Oh, I guess that title sounds a little too dramatic, maybe? As if it’s about some big thing that happened? No, I just want to log some of what I’m learning on this break between day jobs.

Linked, in.

My “professional news” post on LinkedIn from Oct. 4 has done some numbers: 15,000+ views as of today. LinkedIn is my professional social media channel now. It was only a couple years ago that I viewed LinkedIn influencers with a great deal of skepticism, one line at a time, pithy inspirational cliches, etc.

That’s… changed? Somewhat. Although I do deeply enjoy my share of B2B sales and marketing brainrot, I mostly find LinkedIn to be the best place to explore, grow, and develop a professional network, especially as I dip into some new areas of learning and expertise. It’s so easy to follow a bunch of companies and people now, and the algorithm here is… good (?!) at putting valuable content in front of me.

Sustainability. AI. Sustainable AI?

Adding AI to the energy cost of the web — already burdened with cryptocurrency blockchain mining and Web3 — is a disaster, happening now, in not-that-slow motion.

In places like Virginia (where I live), coal is nearly phased out (that’s good), but new natural gas plants are coming online to meet the electricity demand of data centers, where processing power for generative AI is in turn, in demand (that’s bad). And as previously established, there is no end in sight to the data center construction in my personal neighborhood.

That’s mostly bad. But nothing is going to stop the demand for new technology, and although some folks will certainly use AI to optimize the routing and storage of renewable energy (please?), consumptive technology looks like it’s outpacing renewable technology at the moment.

Which is also bad.

So what can we do? Measure the impact, report on it, make the businesses and people using AI tools more aware of the cost of AI in terms of greenhouse gas emissions.

Carbon accounting for AI was just funded last week, with Google Ventures and others putting another $25M into Scope3.com, the somewhat confusingly named (or ambitiously named, depending on your perspective) company that’s previously focused on measuring the carbon emissions of AdTech.

Measuring, reporting, and reducing the most carbon-intensive processes that make our increasingly convenient robot overlords work is the right idea.

Three books

Here’s what I’m reading right now:

• Building Green Software, from O’Reilly: After hearing all three authors and the foreword writer on multiple podcasts, I ordered this from my phone while cooking dinner one night last week, and started reading it by Prius dome-light in a soccer field parking lot a couple days later. (See visual proof below.) A few chapters in, some of it is review of things I’ve gleaned elsewhere, some of it is deeper detail than I had on hand about things like server capacity, energy efficiency, and proportionality, and all of it is useful.
• Never Search Alone, the Phyl Terry book that goes with the community and job search methodology: I’m almost done with a first read of this via the Kindle app on my phone, with lots of secondary PDFs, workbooks, Slack channels, guides, videos, podcasts, etc. riding alongside the book. So far, it’s great. It’s hard to slow down and think (harder than usual) about what I want to do next and how to get there, but I expect it will be worth the effort.
• Personal History, Katherine Graham’s memoir: I did not have “read that Washington Post publisher’s autobiography” on my to-do list, but it’s recommended early in Never Search Alone, and they had a copy at our beloved local public library, so here we are. There are a couple generations of interesting interfaith and interclass marriage and family stories in the first chapters, and I’m half-tempted to go look up her Meyer side’s family tree to see if we’re related. (We’re probably not related.)

What else

When you work at a place that gives you free access to its products (and bandwidth, and storage), sometimes you build up a little bit of eggs-in-one-basket debt you never think you’ll need to undo. My 2021 song-a-day project was almost entirely done on WordPress.com infrastructure, including a (free for me then) paid plan to host… 365 videos, minus a handful posted on YouTube from low-bandwidth vacation locations.

So, I kicked off a fun migration project today to move the video files to YouTube, and I wonder if it’s easier to do an enterprise-grade file migration and search-and-replace operation than it is to do a consumer-grade move, but I’m leaning more about the WP REST API and of course toying with some AI tooling in the process, too.

Did ChatGPT get bored with a long list of song titles today and start hallucinating the wrong file names? Yes. Yes it did. But all of the wrong songs were clearly related to other songs from the list, and in at least one case, it was another song from the list, but in the wrong place. The title that threw it all off? “Hit The Ground Running” by Smog. ChatGPT saw that “Hit The” and, like Roger Rabbit screaming “TWO BITS” said “Road Jack” and then made up four titles in a row.