Podcasts, Google NotebookLM, and my Making Coffee post

As indicated at the beginning of the previous post about making coffee, the post was mounted, definitely tongue-in-cheek, as a result of a podcast created by Google NotebookLM about a previous blog post I’d mounted about making yoghurt muffins.

As part of the initial vision I had when I started to mount the post about making coffee, it was a foregone conclusion that a new podcast be generated as a result. As such, my brother and I tried to make podcasts about the post using Google’s NotebookLM Podcast Generator.

I had a few objectives about how the podcast should be structured:

  • There should be a bent that both the post as well as the podcast were inspired by the previous podcast on yoghurt muffins;
  • There was a hope that the new podcast would be somewhat (read I had an expectation of it being “sufficiently”) self-aware, including that the previous podcast was created “by the same hosts”, and therefore, that “the hosts” would at an obvious juncture make a comment to the effect of “In a previous podcast, we jovially suggested that the blog post author create another blog post about making morning coffee, and whaddya know, they did!”;
  • When supplying Google NotebookLM with source material, I entered an expressed expectation that there be a certain incredulity on the parts of the “hosts” that, albeit tongue-in-cheek, that I followed up on the jovial suggestion that I make such a blog post in the original podcast (at about 3:49), as well as the previous two points.

Unfortunately, the results were less than satisfactory.

  • The first podcast that was created, using the original yoghurt muffins post as one of several pieces of source material, seemed to focus more on the post on the yoghurt muffins than the post on making coffee;
  • Said first podcast generally seemed to rehash the podcast on the yoghurt muffins, often word for word or at least re-phrasings;
  • Throughout several of the podcasts that were created, there seemed to be lot more of what I will call psychobabble that came through multiple times, analyzing myself, and went on tangents on how to look at things differently in life — comments which, in general terms, were not fully inaccurate, but which were general human nature and not really what we were looking for. Ultimately there were increasing departures from the content, even ignoring the meta and self-referential parts.
  • The podcasts became increasingly long, one being at least 15 minutes long.

Ultimately, NotebookLM seemed unable — perhaps by design or a known lack thereof, and certainly at what would constitute a cursory glance — to handle levity, self-awareness, and meta conversations / commentary.

Anyway, here is the podcast I settled on after a few tries and was frustrated with the general results; it at least does discuss somewhat the meta nature of the post, as well as just barely enough of the coffee part when taken at face value.

Sigh. (Meaning, I guess that if I want a podcast creator that does what I want, I should find out if such exists. Or, I dunno, make a real life podcast myself.)

Google Maps seems to need to learn that some streets go East AND West

I think that Google Maps is overlooking a basic function: In the real world, people sometimes go east, and sometimes go west.

Yesterday for the third time in a couple of years I relied upon Google Maps for directions and was sent to the wrong place. Caveat Emptor strikes again.

In Montreal, east-west streets which bisect St. Laurent Boulevard (which, no surprise, goes sort of north-south), start their numbering in both east and west directions from there. Hence you can have two equally valid addresses on a given street, given the proviso that one is designated as “East” and the other “West”. (Hey! It’s Captain Obvious!)

Fortunately, the address I was looking for was 151; during an hour of going around the neighbourhood looking for parking around “151 Laurier” (East as proposed by Google Maps), I found out that that address wasn’t a dépanneur that sells a huge variety of microbrewery beers, and looked like it never was, and finally decided to go further down the street looking for similar businesses. I suddenly had a V-8 moment and realized “Ooops what about 151 Laurier WEST?” I high-tailed it in the opposite direction and found the business in question. And to my disappointment, they were out of the particular beer I was seeking — Weizenbock, by La Brasserie Les Trois Mousquetaires, which has replaced my previous definition of ambrosia, Trois Pistoles by Unibroue.

Twice before I have had similar experiences:

About a year ago, while in Western Canada in completely unfamiliar territory on a business trip, I had looked up a client’s address, and not knowing about any local east/west splits that addresses on the Trans-Canada Highway may have in that locality, I tried to find the address, on the east end of town, that Google Maps had provided; I was about 45 minutes late by the time I finally managed to suspect that my client’s address was a “West” address and got there.

And just to quash any participant in the Peanut Gallery out there about to say “Aha well when using Google Maps you should know that in such cases they’ll always send you to the East address, so be sure to always check both!” a couple of years ago I had looked up a local address for client, and Google sent me to Gouin Boulevard West here in Montreal, a solid 45 minute drive away from my client’s Gouin Boulevard East address.

Now the Peanut Gallery may have a point: In the real world, people sometimes go east, and sometimes go west. And when it comes to using a free online service, you get what you paid for. As such, when looking up an address on any online service, one should notice “Hmmm this is an east-west street which may bisect such and such a street and as such have East addresses and West addresses; I should specify both east and west in my address search.”

But I wonder how many other people place enough faith in Google that under such circumstances — such as when they don’t know that there’s an East and West of a given street — they would reasonably expect in the case that a street has valid East addresses and valid West addresses (and likewise for North and South addresses) that Google’s response page would come back with “Did you mean (A) 151 Laurier East, or did you mean (B) 151 Laurier West?” Certainly Google seems good enough at asking such a question when you slightly misspell a street or city name, or decides that it doesn’t recognize the address you supply and provide you with half a dozen options, as often spread across the country as spread across the city.