How I got Amazon Prime almost for free

Last summer, my appetite for quantifying my finances intersected with my food-procurement habits to yield a math exercise: How much of my Amazon Prime membership was I chipping away with these discounts at Whole Foods?

The Seattle retail leviathan’s 2017 purchase of the Austin-based grocery chain consolidated a large portion of my annual consumer spend at one company. It also gave me a new set of benefits for the Amazon Prime membership my wife and I have had since 2011: an extra 10% off sale items except beer and wine, plus some Prime-only deals.

(Personal-finance FYI: Amazon also touts getting 5% cash back at Whole Foods on its credit card, but the American Express Blue Cash Preferred offers 6% back on all grocery stores. That higher rate combined with Amex Offers for rebates at designated merchants easily erases the card’s $95 annual fee and returns more money than I’d get from Amazon’s card.)

So on my way out of Whole Foods, I created a new Google Docs spreadsheet on my phone and jotted down the Prime savings called out on my receipt. Then I did the same thing after subsequent visits. If Whole Foods and Amazon were going to track my shopping habits (which I assume they could from seeing the same credit card even if I didn’t scan in the QR code in the Amazon app at the checkout), I ought to do likewise.

Aside from $10-and-change savings during last July’s Prime Day promotion and again on roses for Valentine’s Day, most of these 41 transactions yielded $4 or less in Prime discounts. But after a year, they added up to $118.14, just 86 cents less than the $119 Prime annual fee.

To answer the obvious question: No, I did not step up my Whole Foods visits because of this tie-in. That place does happen to be the closest almost-full-spectrum grocery store to my home, but there’s a Trader Joe’s barely further away that trades a smaller selection for cheaper pricing on staples like milk and flour. And thanks to this dorky habit of mine, I can tell that I’ve shifted more of my business from WF to TJ’s the past few months.

Lessons from moderating three virtual panels

I spoke at a conference Tuesday for the first time since February, and this time the dress code was a little different: no pants required.

That’s because my appearance at Futureproof IT came through my laptop’s webcam and the Zoom video-chat app, leaving nothing below my chest visible to the remote audience. That’s also how I moderated two more panels this week for the Collision conference that was originally set to run in Toronto in June but has since migrated to a digital format, with my appearances among those recorded in advance.

And that’s how I expect all of my conference speaking to happen for at least the next few months, thanks to the novel-coronavirus pandemic ruining everything.

I’ve learned a lot about successful panel moderation on a physical stage, but doing so in a virtual environment brings new challenges.

Start with picking and positioning a webcam. The camera in my aging iMac is at a good height relative to me when sitting, but it also delivers a subpar 720p resolution–and from that angle, I’d have natural light from the windows hitting only one side of my face. My HP laptop, meanwhile, has a 1080p webcam, and by parking it atop a stepladder and then a large tin topped by the thickest book I could find in my office (a hardcover of Dune), I could position it high enough while allowing myself to face my office’s windows.

(If I’d only bought the Wirecutter-endorsed Logitech C920S webcam in the Before Times, I could have stuck the thing on a tripod and be done with it. As the song goes, there’s a lot of things if I could I’d rearrange.)

The Collision panels added another complication: a request for a pale, blank backdrop. I managed that by hanging a white bedsheet from the ceiling with packing tape and binder clips–the tape stuck to the drywall, but I needed the clips to hold the tape to the sheet.

And then none of my other panelists showed up with pale, blank backgrounds. That’s one reassuring aspect of this: Not only can you expect somebody else to have audio or video hiccups, you can also expect somebody else to have a worse backdrop or camera angle.

Before kicking off a virtual panel, you must also silence every other device in the room, and my failure to think through “every other device” meant the Futureproof panel was not interruption-proof. As in, we were distracted by the one thing I didn’t think to put in do-not-disturb mode, an old Trimline land-line phone on my desk mainly for nostalgia purposes.

At least I didn’t have to change any settings on the laptop, thanks to Windows 10’s Focus Assist quashing interruptions from other apps once I switched Zoom to full-screen mode. But that also meant I had to look elsewhere for a timer: I couldn’t see the lock in the Windows taskbar, while Zoom’s option to show your connected time doesn’t account for minutes spent prepping on a call before a panel begins. I made do with the clock apps on my phone and iPad.

All three panels suffered from a certain latency as other speakers paused before answering my questions. You can’t point or nod to one as you would onstage to encourage them to jump in–and if one starts filibustering, it’s also harder to signal him to wrap things up. Simply reading the facial expressions of other panelists can be difficult if they use a lower-resolution webcam or neglect their lighting.

Reading the audience seems even harder in Zoom unless, I guess, you keep the chat pane open and have an audience that is exceptionally concise in their feedback. The way Facebook and Twitter let a live video audience respond with emoji and hearts ought to deliver easily-understood feedback at scale, and perhaps one of the many virtual-event apps now seeing escalating interest–see my friend Robin Raskin’s writeup of a handful at Techonomy–gets even closer to the real thing.

But I highly doubt any app will recreate how great it can feel to have a live audience tuned into the talk, laughing at your jokes and then applauding your work.

A new bread-baking hack: tangzhong

I’ve been baking sandwich bread about every week since 2004, and two weeks ago I learned that I’d been missing out on a fairly simple technique to make softer loaves that stay fresh longer.

It’s called tangzhong, and I learned about it from a tweet from science journalist Erin Biba. I read the introduction she shared from King Arthur Flour’s blog, then that blog’s explanation of how to work this into a standard sandwich-bread recipe.

“Tangzhong” is either Chinese for “soup starter” or “flour roux,” depending on where you read, and it’s a way to incorporate more water into bread dough without it evaporating and making the bread stale as fast as usual. You do that by cooking a small amount of flour and water in a pan until they form a slurry–as if you were making roux for homemade mac and cheese–that locks in the moisture, Then you combine this with  the rest of your flour mixture before adding your yeast mixture and proceeding as usual.

The King Arthur directions started simple: 3 tablespoons flour and half a cup of water in a pan over medium heat until the mixture thickens. But then math intruded, in the form of calculations to determine how much water to add to the rest of the recipe to ensure that the water added up to 75 percent of the weight of the flour.

The first try, I spent so much time weighing ingredients–a somewhat irritating step when you’re putting flour and then water on a scale in measuring cups already labeled to tell you how much of something they contain–that I forgot to add the weight of that initial half cup of water back in and instead poured in almost a quarter of a cup more water into my yeast-and-water mix. The result was an unmanageably soupy dough that I couldn’t work with until I’d added maybe another half cup of flour above the 3 1/2 cups in my standard recipe.

That left me with more risen dough than I’d need for one loaf, so I broke off a bit and formed that into hot dog rolls and hamburger buns. All were delicious, if done at least half an hour behind schedule. The bread was notably softer and fluffier–more like the pre-packaged sort I renounced buying 15 years ago when I was just embarking on baking hipsterdom–and kept fresh longer, freeing me from wanting to relegate the last couple of slices to toast or a grilled-cheese sandwich.

The second try, I decided to wing it a little and not add any water beyond the half cup in the tangzhong and the 1 1/4 cup in which I normally pour yeast and honey to start. I also saved myself some complexity by putting all of the flour in the stand mixer’s bowl, then taking out 3 tablespoons for the tangzhong. But once again, I still had dough too damp to work, so I added a little more flour and once again got some bonus rolls.

On the third try, I used only 1 3/4 cup of water to start the yeast, leaving me with 1 1/2 1 1/4 cups of water in the total recipe. As in, close enough to not too far below the proportions the King Arthur recipe had specified, if only I’d paid more attention the first time, but also the same amount of total water as in my usual bread-baking practice.

On this iteration, the dough came away from the sides (if not the bottom) of our stand mixer like usual, and the finished loaf was fantastic like the others. As I write this, I am already looking forward to tomorrow’s sandwich for lunch.

Updated 5/16 to correct totals for the third iteration.

Here’s why I have trouble buying things quickly online

Like many of you, I’ve been doing a fair amount of online shopping. But I’ve probably been much slower at it than most of you.

Not “slower” in the sense of taking forever to pick one product over another (although my indecision-making there is considerable), but in the sense of deciding how I’ll pay for it and which third-party site I should click through before making the purchase.

Picking a credit card is the easier part even if I’m not buying stuff for my job (work expenses go on a separate card to ease my accounting). Most of the time, the 2% cash-back rebate on the Citi Double Cash Mastercard makes it the obvious choice. It’s been an even easier call when Citi’s offered extra cash back in promotions with various merchants.

But other card issuers have their own extra incentives. American Express and Chase offer extra cash back and do so much more often, but you have to sign up for each such offer on their sites or in their mobile apps. So I need to consult both before any purchase–and then hope the merchant in question doesn’t drop one of these deals the week after my purchase.

(Note that Amex and Chase also have tiered cash and points rewards for categories outside of online retail; a proper discussion of them would require a separate post.)

Not too many years ago, my shopping decisions would have ended there. But then I had to start considering shopping portals, the points-for-purchases sites most frequent-travel programs provide for members. By itself, a mile or a point for a dollar spent somewhere is barely worth thinking about. But that incremental addition does deserve your time if you’re nearing an award-redemption threshold or have miles or points that will expire without new activity in your account.

These portals don’t all offer the same rate or each offer the same rate over time. To verify which one offers the most return, I use a site called Cashback Monitor that tracks these deals and lets you set up a custom page with your favorites. (For more details, see this concise how-to by One Mile At A Time’s Tiffany Funk.) JetBlue consistently offers some of the best earn rates; fortunately, TrueBlue points have not seem the same deflation as other frequent-flyer currencies in recent years.

You may find no future-travel benefit for a potential purchase. Best Buy, Target and Walmart recently seem to have dropped out, while I can’t remember seeing any travel incentives for Amazon. In those cases, I’ll go to one last site before starting a purchase: my client Wirecutter, which often tells me what to buy and makes a decent chunk of its money off affiliate payments from Amazon and other retailers. If I can’t treat myself to a little kickback on a purchase, helping one of my favorite clients seems like a decent fallback.

The quesadilla assembly line

I’m in the middle of my longest stretch of home cooking in years, so I’m falling back on one of the recipes I always make before heading out of town for work.

Quesadillas check off several critical requirements in family cuisine: cheap ingredients, simple to prepare, most picky-eating kids will eat them, most grownups like them too, easily reheated, freezer-tolerant.

I won’t call this recipe authentic; I can’t object if you label it cultural appropriation, given my absence of Latino heritage. It is, however, an affordable and low-stress way to put together dinner for multiple nights. So I hope this helps some of you trying to avoid total dependence on takeout and delivery.

  • 1/3 onion
  • 1/2 bell pepper
  • 2 tbsp. cooking oil
  • 1/4 tsp. salt
  • 1/4 tsp. cumin
  • 1/4 tsp. chili powder
  • 1 15-oz. can black beans
  • 1 ripened avocado
  • 1/2 lb. Monterey jack cheese
  • 10 8-in. flour tortillas

Dice onion and bell pepper. Warm 1 tablespoon oil in a pan over medium heat, then sauté the vegetables until softened while adding the salt, cumin and chili powder.

Slice the avocado in half and remove the pit, drain and rinse the black beans, and divide the cheese into 10 portions. Place a nonstick pan or griddle on the stovetop over no more than medium heat.

Put a drop of vegetable oil on aluminum foil or another clean and washable or disposable surface and swab it with one side of a tortilla. Take one heaping tablespoon of avocado and mash it on half of the other side of the tortilla. Add a dangerously-heaping tablespoon of black beans atop the avocado, then a tenth of the sauteed veggies. Dice one of the one-tenth portions of the cheese and scatter that over the other fillings.

(This is also your opportunity to add any other random ingredients that could suit this context: a tomato diced up, leftover BBQ, chopped spinach or arugula from the garden or remaining from a grocery-store purchase, diced cilantro, etc.)

Fold the quesadilla over into a semicircle, then cook it on the pan or griddle until browned on each side while frequently pressing down on it with a spatula to ensure the ingredients meld. You should be able to do at least two at once, but move them often to avoid them lingering on any cool spots in the pan or griddle.

Serve as is or with salsa or sour cream. Try to save some of the batch for leftovers.

Things I learned from working a primary election

After more than 15 years of writing about voting-machine security, I finally got some hands-on experience in the field–by waking up at 4 a.m. and working a 16-hour day.

I’d had the idea in my head for a while, thanks to frequent reminders from such election-security experts as Georgetown Law’s Matt Blaze that the best way to learn how elections work is to work one yourself. And I finally realized in January that I’d be in town for the March 3 Democratic primary and, as a self-employed type, could take the whole day off.

I applied at Arlington’s site by filling out a short form, and about two hours later got a confirmation of my appointment as an election officer. (My wife works for Arlington’s Department of Technology Services but has no role in election administration.) A training class Feb. 11 outlined the basics of the work and sent me home with a thick binder of documentation–yes, I actually read it–and on March 3, I woke up two minutes before my 4 a.m. alarm.

After packing myself a lunch and snacks, as if I were going to grade school, and powering through some cereal, I arrived at my assigned polling place just before the instructed start time of 5 a.m. I left a little before 9 p.m. Here are the big things I learned over those 16 hours:

  • Yes, having people fill out paper ballots and scan them in works. I saw 500-plus voters do that while I tended the scanner in the morning, and none had the machine reject their ballot. There was confusion over which way to insert that ballot, but the scanner accommodated that by reading them whether they were inserted upside down, right-side up, forwards or backwards. (I wish more machines were that tolerant of human variances in input.) And at the end of the day, we had a box full of ballots that will be kept for a year.
  • The technology overall appeared to be of higher quality than the grotesquely insecure, Windows-based Winvote touchscreen machines on which I voted for too many years. This scanner was an offline model running a build of Linux, while the poll-book apps ran on a set of iPads.
  • The “vote fraud” rationale for imposing photo ID requirements is not only fraudulent, but photo IDs themselves are overrated. The state allows a really broad selection of public- and private-sector IDs—unavoidable unless you want to make it obvious that you’re restricting the franchise to older and wealthier voters—and our instructions required us to be liberal in accepting those. I didn’t see or hear of anybody getting rejected for an ID mismatch. (The one surprise was how many people showed up with passports; I quickly grew to appreciate their larger color photos over the tiny black-and-white thumbnails on drivers’ licenses.)
  • Asking people to state their name and address, then matching that against voter-registration records, does work. That also happens to be how voter check-in used to work in Virginia before Republicans in the General Assembly shoved through the photo-ID requirement that’s now been reversed by the new Democratic majority in Richmond.
  • You know who really loves high turnout? Election officers who otherwise have some pretty dull hours in mid-morning and then mid-afternoon. At one point, the person in charge of the ballot scanner busied himself by arranging stickers into a bitmapped outline of Virginia, then added a layer of stickers on top of that to represent I-95 and I-66. Fortunately, precinct 44 blew away past primary-turnout records with a total of 1,046 in-person votes.
  • The attention to detail I saw was almost liturgical. Every hour, the precinct chief did a count of voters checked in and votes cast to ensure the numbers matched; every record was done in at least duplicate; every piece of paper was signed by at least two election officers, and the overall SOR (statement of results) bore the signatures of all eight of us. We closed out the night by putting documents and records in specified, numbered envelopes, each locked with a numbered zip-tie lock; each number was recorded on a piece of paper on the outside of each envelope that was itself signed by two election officers.
  • Serving as an election officer isn’t physically demanding work, but it does make for a long day. We did have coffee delivered, but it didn’t arrive until 9 a.m., and nobody had time for dinner during the rush to close out things after the polls closed.
  • It’s also not the most lucrative work ever. My paycheck arrived Friday: $175, amounting to an hourly wage of $10.94. The value of seeing the attention paid to make democracy work and then watching more than a thousand people show up to exercise their rights: priceless.

Updated 3/23/2020 to fix some formatting glitches.

Work-from-home advice from a work-from-home regular

My occupational routine of working from home is suddenly in fashion for the dreadful reason of a global pandemic. Employers ranging from Google to the federal government to the Washington Post have been telling people to get out of the office and stay out until some sort of all-clear is declared about the novel coronavirus.

This may be a new and unsettling development to many of you, but it’s been my everyday reality for the past nine years–longer, if you count all the time I’d work from home while at the Post to test one gadget or another.

The joking on Twitter that “the only ones to survive will be freelance writers” may overstate things a bit, but all of this Me Time has left me well versed at staying productive without such traditional work delineations as a commute to a geographically distinct workplace and frequent in-person professional interaction with other human beings.

Here are the best practices I’ve learned since 2011 or so:

  • Have a spot at home that serves as your logical office. Ideally it’s a physically separate room–if you’re self-employed, the home-office deduction is easier to claim that way–but it should be someplace you can associate with work. And can then leave when you’re not on the clock.
  • Get a comfortable chair (I should have followed this advice years ago instead of letting my current chair get even more worn out) and make sure it’s positioned so you can type comfortably for hours at a stretch
  • You don’t need a separate webcam–unless your laptop has one below the screen that treats video callers to an up-nostril perspective of you–but a desktop USB microphone would be a good idea. My client Wirecutter has some useful advice; you should be fine with the budget pick unless you do podcasts for a living.
  • Make sure that your webcam shows a tidy office to the rest of the world. You can still have piles of paper and dirty clothes around; just keep them out of the frame.
  • You will probably spend a lot more time on conference calls, and some con-call systems are more evil and stupid than others. Please try to lead your office away from the ones that date to 1980s telecom and and to apps like Zoom or Uberconference that indicate who’s speaking at any time. Note that the free version of Zoom limits meetings to 40 minutes, which is such a good reason not to pay that I must wonder if this company is trying to go out of business.
  • Does your WiFi offer reliable coverage in your home office? If it doesn’t, you will notice that intensely and often once you’re clocking eight hours a day on that questionable connectivity. And no matter what, you should have all of your important documents cached or copied for offline access.
  • You should know what kinds of backup bandwidth are available–for example, major cable operators say they will open their WiFi hotspot networks to the public, while Sprint and T-Mobile plan to offer their subscribers 20 GB of mobile-hotspot usage.
  • Yes, you still need to shower and get dressed. But you may find that you can use those daily habits as fake deadlines: No showering until I finish this task that I didn’t get done yesterday.
  • Find ways to shut out distractions. If you find yourself wandering down Wikipedia rabbit holes, clean part of your house instead. Or go outside and get in some gardening, if it’s warm enough. If nothing else, walk around pointlessly your home as you would in an office.
  • We all have coworkers who don’t reply to e-mails fast enough. Figure out what comms channel works to bug them when they inevitably leave your last message unanswered: Slack, a text, a call, a direct message on their most common social platform.
  • Don’t eat lunch at your desk. Ever. You’re at home, and you don’t have to do that anymore. While you’re at it, get in the habit of making yourself lunch; you can put the savings into patronizing the restaurants, coffee shops and bars closest to you.
  • It’s okay to run short errands during the day. It’s not like you were that productive over every hour of your in-office workdays anyway.
  • Get to know your neighbors, especially those who have been working from home all along and who may have useful neighborhood-specific advice. Human contact during the day is good.
  • You’ll also soon realize which of your neighbors insist on hiring people to tidy up their yard with noisy, polluting gas-powered leaf blowers.
  • Have some kind of back channel–a text or WhatsApp group, a Facebook Messenger group, a Slack channel, whatever–for personal banter with your favorite fellow cubicle-farm dwellers.
  • Take time to call friends about absolutely nothing.
  • You can swear at your computer as much as it deserves without freaking out co-workers, but please don’t get in that habit anyway. (This is literally me saying “do as I say, not as I do.”) Especially if you’ve got a kid stuck at home too.
  • On the other hand, go ahead and play your preferred productivity playlist through your computer’s speakers. If blasting Kool Moe Dee’s “I Go To Work” or R.E.M.’s “Finest Worksong” gets your day in gear, you don’t need to confine that to headphones. (This is totally me showing my age.)
  • If you’re tired, you’re allowed to nap. You’re at home! Nobody outside can tell you’re enjoying a postprandial snooze.

(My thanks to everybody who replied with further suggestions to the Twitter thread in which I first shared most of these tips.)

Updated 3/18/2020 with a few extra tips.

Here’s how much Facebook was tracking me around the rest of the Web

Facebook finally fulfilled one of Mark Zuckerberg’s campaign promises this week–a promise dating back to May of 2018.

That’s when Facebook’s CEO said the company would roll out a “Clear History” feature that would let its users erase Facebook records of their activity at other sites and apps as gathered by the social network’s Like and Share buttons and other plug-ins.

(If it took you a long time to realize the extent of that tracking, I can’t blame you. Instead, I can blame me: The post I did for the Washington Post when my old shop integrated a batch of Facebook components to its site didn’t spell out this risk.)

Twenty months after Zuck’s announcement, this feature, renamed “Off Facebook Activity”, finally arrived for U.S. users on Tuesday. I promptly set aside that day’s tasks to check it out firsthand.

The good news, such as it was: Only 74 apps and sites had been providing Facebook info about my activity there. And most of them (disclosure: including such current and past clients as USA Today, Fast Company, The Points Guy and the Columbia Journalism Review) had only coughed up isolated data points.

The bad news: The Yelp, Eventbrite, AnyDo, and Duolingo apps had all coughed up more than 20 records of my interactions there, as had the sites of Home Depot and Safeway owner Albertson’s.

To judge from the responses I got from readers of my Facebook when I asked them how many sites and apps showed up in their own Off-Facebook Activity listings, I’m practically living a cloistered life. Most comments cited three-digit numbers, two close to four digits: 232, 356, 395, 862, and 974. One thing most of these users seemed to have in common: using Google’s Chrome as a default browser instead of Apple’s Safari or Mozilla Firefox, both of which automatically block tracking by social networks on other pages. The former is the default on my desktop, while the latter has that place on my laptop.

I’ve now cleared my history and turned off future Off-Facebook Activity–at the possible cost of no longer having WordPress.com publish new posts automatically to my Facebook page. I can probably live with that.

The boring art of testing hotspot bandwidth and battery life

I’m nearing the finish line (I hope) of an overdue update to the Wirecutter guide to WiFi hotspots. The research for that had me repeatedly subjecting an array of loaner hotspots from all four nationwide wireless carriers to tests of the two core metrics of bandwidth and battery life.

It hasn’t exactly been my most exciting work.

For bandwidth testing, I’ve continued to rely on Ookla’s Speedtest.net Web, Android and iOS apps to clock the download speeds, upload speeds and ping times each hotspot has served up. This is pretty much an industry-standard benchmark, and these apps are simple enough to run.

But getting data out of them is another thing:

  • The iOS app creates a .csv file you can open in any spreadsheet app that includes every relevant bit of data–date and time, GPS-derived location, WiFi network name, download/upload/ping measurements, shareable public link–and attaches it to an e-mail message.
  • The Android app also generates a .csv file–except that choosing to have it saved to your Google Drive leaves you with a .eml mail-attachment file. You have to e-mail it to yourself to get a usable .csv, at which point you discover that this export doesn’t include the name of the wireless network.
  • The Web app’s “Export” button yields a third type of .csv file, one without a record of the WiFi network name, your location, or a shareable link.
  • No, the option to create a Speedtest account won’t help–because you can’t log into that from the mobile apps.

Ookla is owned by PCMag publisher Ziff Davis, but that has yet to result in any corporate pressure to make exporting measurements less janky for the hardworking journalists at that and other Ziff tech media properties.

Testing hotspot battery life only requires recording the times you started and ended each trial. But because these things often run for 12 hours or more, it’s not realistic to tether a laptop to a hotspot and keep working nonstop until the hotspot battery expires and the connection drops.

To ensure my laptop would be keeping each hotspot working full time, I opened a page to NASA’s live YouTube channel. Beyond running up the social-media metrics for one of my favorite four-letter government agencies, keeping the browser on a single live channel avoids the risk of YouTube’s recommendations sending me off to some nutcase conspiracy hub.

Because I’m not always that smart, I didn’t think to check my laptop’s ability to log a wireless connection going offline until after I’d spent an hour and change watching one hotspot linger at 1% of a charge.

As a helpful StackExchange thread pointed out, that logged data awaits inside Windows. Type “Event Viewer” in the taskbar search, open that app, select “Applications and Services Logs” in the left-hand pane, double-click the center pane’s “Microsoft,” “Windows,” “UniversalTelemetryClient,” and “Operational” entries in succession, then select “Filter Current Log…” in the right-hand pane. Type “55” in the resulting dialog’s Event ID field, hit “OK” and you’ll see a series of entries.

Assuming you check this right after seeing that the laptop went offline, opening the most recent should reveal a properties field consisting of “Is the Internet available: false,” with the time corresponding to when the hotspot died.

Since I don’t have a Mac laptop, I’m not sure how you’d do this on one. A different StackExchange thread suggests a Terminal command, but that doesn’t work on my iMac–maybe because this aging desktop isn’t running the latest macOS edition. It would be ironic if you have to hit the command line on a Mac to perform a task that Windows lets you accomplish inside a graphical user interface–but the Windows Event Viewer app is mighty ugly itself, and neither operating system covers itself in glory in this aspect.

Google told me to put this blog on a diet

The screen real estate around these words should look a little neater now–not much thanks to my own editing instincts. Instead, I needed a scolding from Google.

That came from its PageSpeed Insights tool, a page you can use to check the performance of your own site or any other. It’s been around for years, but I didn’t think to use it until after I finally connected this site to Google’s Search Console webmaster tool last May (yes, even though I’ve been blogging here since 2011 and am supposed to be a professional tech journalist) and saw a prompt to try PageSpeed Insights in my results.

The results were not flattering: a mobile performance score of 47 out of 100, with desktop performance better but still subpar at 89. Scrolling down revealed a variety of flaws in this blog, some that I could correct with the limited tools of a free WordPress.com account and some that I could not.

I started with the first obviously fixable thing, image sizes. Google suggested that I convert such frequently-downloaded elements as the background image to “next-gen formats” like WebP or JPEG 2000. But instead of switching to files that Apple’s Safari browser can’t display, I opened the original photo for the background (a view of the Blue Ridge from a bike century ride in 2006) in my Mac’s GraphicConverter app and exported it with a lower image-quality setting that cut its size by a third at no visible cost.

That did not impress PageSpeed–my mobile score actually dropped to 44–so I moved on to the third-party code that Google reported had been getting in the way of my own content. It listed Facebook as a major offender, accounting for 207 milliseconds of delay. Removing the Facebook widget that you formerly saw in the right-hand sidebar boosted my mobile score to 53, with the desktop score essentially unchanged at 87.

(PageSpeed Insights scores fluctuate up and down with each test, so don’t get bent out of shape if yours drops by five for no apparent reason, at least not until you run the test again.)

Then I removed the Facebook and Twitter share buttons and elected to toss all of the share buttons except the tool to e-mail a link to the page. That edged the mobile score up to 58 and pushed the desktop score up to 96.

This pruning got me to look again at all the ads you see around here–and think about how low-rent so many of them are. I turned off the three “additional ad placement” options that had been salting each post with extra banners. That may cut into my ad revenue, but it’s already so thin I’d rather that my professional presence online not look quite so janky.

Having spun myself up fully into Marie Kondo mode, I returned to the sidebar and removed the Twitter widget (PageSpeed Insights had blamed Twitter code for 119 ms of delay) and some non-interactive links that only cluttered that part of the page.

The results of all this effort as calculated by PageSpeed Insights just now: scores of 76 on mobile and 99 on desktop. I hope you notice this blog loading a little quicker–and I trust you appreciate how this exercise has also rid my blog of Facebook’s tracking.