MapScaping Podcast: Pg_EventServ

Last month I got to record a couple podcast episodes with the MapScaping Podcast’s Daniel O’Donohue. One of them was on the benefits and many pitfalls of putting rasters into a relational database, and the other was about real-time events and pushing data change information out to web clients!

TL;DR: geospatial data tends to be more “visible” to end user clients, so communicating change to multiple clients in real time can be useful for “common operating” situations.

I also recorded a presentation about pg_eventserv for PostGIS Day 2022.

Keynote @ CUGOS Spring Fling

Last month I was invited to give a keynote talk at the CUGOS Spring Fling, a delightful gathering of “Cascadia Users of Open Source GIS” in Seattle. I have been speaking about open source economics at FOSS4G conferences more-or-less every two years, since 2009, and took this opportunity to somewhat revisit the topics of my 2019 FOSS4GNA keynote.

If you liked the video and want to use the materials, the slides are available here under CC BY.

MapScaping Podcast: Rasters and PostGIS

Last month I got to record a couple podcast episodes with the MapScaping Podcast’s Daniel O’Donohue. One of them was on the benefits and many pitfalls of putting rasters into a relational database, and it is online now!

TL;DR: most people think “put it in a database” is a magic recipe for: faster performance, infinite scalability, and easy management.

Where the database is replacing a pile of CSV files, this is probably true.

Where the database is replacing a collection of GeoTIFF imagery files, it is probably false. Raster in the database will be slower, will take up more space, and be very annoying to manage.

So why do it? Start with a default, “don’t!”, and then evaluate from there.

For some non-visual raster data, and use cases that involve enriching vectors from raster sources, having the raster co-located with the vectors in the database can make working with it more convenient. It will still be slower than direct access, and it will still be painful to manage, but it allows use of SQL as a query language, which can give you a lot more flexibility to explore the solution space than a purpose built data access script might.

There’s some other interesting tweaks around storing the actual raster data outside the database and querying it from within, that I think are the future of “raster in (not really in) the database”, listen to the episode to learn more!

LLM Use Case

I can only imagine how much AI large language model generated junk there is on the internet already, but I have now personally found one in my blog comments. It’s worth pointing out, since comment link spam is a long time scourge of web publishing, and the new technology makes it just that little extra bit invisible.

The target blog post is this one from the late 2000’s oil price spike. A brief post about how transportation costs tie into real estate desirability. (Possible modern day tie in: will the rise of EVs and decoupling of daily transport costs from oil prices result in a suburban rennaisance? God I hope not.)

The LLM spam by “Liam Hawkins” is elegant in its simplicity.

Blog Spam

I imagine the prompt is nothing more complex than “download this page and generate 20 words that are a reasonable comment on it”. The link goes to a Brisbane bathroom renovation company, that I am sure does sterling work and maybe should concentrate on word of mouth rather than SEO.

I need to check my comment settings, since the simplest solution is surely to just disallow links in comments. An unfortunate degredation in the whole point of the “web”, but apparently necessary in these troubled times.

My Subscriptions

It is the age of the unbundled subscription, and I am wondering how long it will last? And also, what do our subscriptions say about us?

Here are mine in approximate order of acquisition:

  • New Yorker Magazine, I have been a New Yorker subscriber for a very long time, and for a period in my life it was almost the only thing I read. I would read one cover-to-cover and by the time I had finished, the next would be in the mail box, and the cycle would repeat.
  • Amazon Prime, I was 50/50 on this one until the video was added, and then I was fully hooked. It’s pricey, and intermittently has things I want to watch, so I often flirt with cancelling, but not so far.
  • Netflix, for a while this was too cheap to not get, the kids liked some of it, I liked some, there were movies I enjoyed. However, the quality of is going down and the price up so it might be my first streamer cancellation.
  • Washington Post, I got lucky and snagged a huge deal for international subscribers which has since disappeared, but got me a $2 / month subscription I couldn’t say “no” to, because I do read a lot of WP content.
  • Talking Points Memo, the best independent political journalism site, which was pivoting to subscription years before it became cool. My first political read of every day.
  • The New York Times, a very pricey pleasure, but I found myself consuming a lot of NYT content, and finally felt I just had to buck up.
  • Disney+, for my son who was dying to see all the Star Wars and Marvel content. Now that he’s watched it all, we are discovering some of their other offerings, they own a quality catalogue.
  • Spotify, once the kids were old enough to have smart phones, the demand for Spotify was pretty immediate. I’ve enjoyed having access to this huge pile of music too (BNL forever!).
  • Slow Boring / Matt Yglesias, my first sub-stack subscription. You can tell a lot about my political valence from this.
  • Volts / David Roberts, highly highly recommended if you are a climate policy nerd, as he covers climate and energy transition from every angle. Never easy, never simplistic, always worth the time.

In the pre-internet days I was also a subscriber to Harper’s and The Atlantic, but dropped both subscriptions some time ago. The articles in Harper’s weren’t grabbing me.

The real tragedy was The Atlantic, which would publish something I really wanted to read less than once a month, so I would end up … reading it on the internet for free. The incentive structure for internet content is pretty relentless in terms of requiring new material very very frequently, and a monthly publication like The Atlantic fits that model quite poorly.

Except for Volts, this list of paid subscriptions is curiously devoid of a huge category of my media consumption: podcasts. I listen to Ezra Klein, Chris Hayes, Strict Scrutiny, Mike Duncan, and Odd Lots for hours a week, for free. This feels… off kilter.

Although I guess a some of these podcasts are brand embassadors for larger organizations (NYT, NBC, Bloomberg), it seems hard to believe advertising is really the best model, particularly for someone like Mike Duncan who has established a pretty big following.

(If Mike Duncan committed to another multi-year history project, I’d sign up!)

One thing I haven’t done yet is tot up all these pieces and see how they compare to my pre-internet subscription / media consumption bill. A weekend newspaper or two every week. Cable television. The three current affairs magazines. The weekly video rental. Even taken ala carte, I bet the old fashioned way of buying did add up over the course of a year.

I’m looking forward to a little more consolidation, particularly in the individual creator category. Someone will crack the “flexible bundle” problem to create the “virtual news magazine” eventually, and I’m looking forward to that.