Earth Notes: On Website Technicals (2020-01)Updated 2022-10-23.
2020-01-31: Frugality Section
To help better group articles with common themes and drivers, and to simplify the site guide, EOU now has sections for: datasets, frugality, microgeneration, podcast, research. Being frugal (ie conserving) is the newest of the bunch.
2020-01-19: Hashtag Magic
As of today, any #hashtag in the article description is automatically linked through to a Twitter search for that tag, except for 'lite' pages.
For posterity, this is what the newsflash banner looks like as of today:
FLASH: Radbot is fundraising - join me and nearly 500 investors, cutting carbon and bills!
2021-01-24: for reference, here's the banner that I'm running a year later. Note the padding to avoid "tap target too small" complaints from Lighthouse/GSC.
2020-01-12: Adsense Funnies
In AdSense, I have now managed to block AutoAds from all AMP and m-dot pages. I have also been experimenting with ad load, though generally keep it under the mid/default position for EOU.
I had to temporarily turn off AutoAds entirely as for example AutoAds was taking the very top ~30% of the mobile window for an ad, and then leaving it blank (when browsing the www site with Chrome on my Fairphone 3). Preview was not reflecting this, though AutoAds may have been reacting badly to my blocking lots of other greedy ATF (Above The Fold) potential ad spots.
Essentially anything before the headline and first paragraph or so of real text content seems pushy to me, and would annoy me as a visitor. AdSense AutoAds has no such reservation, it seems.
Interestingly, it looks as if peak search clicks day for EOU as reported on GSC for the last 16 months was mid-November, with a nadir on Christmas Day then fairly good rebound through mid-January.
2020-01-11: GSC Speed Report Funnies
The bizarre switching back and forth between 'fast' and 'moderate' of all the mobile URLs in the experimental GSC speed report has continued.
I have further restricted the number of AMP pages that will host any ads at all (to just the most popular and recently edited) to maximise the chance of GSC's sample set having a low enough average FCP (First Contentful Paint) metric. The set used in GSC's sample is not clear.
I also note that the FCP times reported in GSC (hovering either side of 1.0s) are about half of those reported to by the linked Pagespeed Insights tool...