This past weekend I was reminded that Phoenix is largely what it is today -- the 5th biggest city in the US -- because of two very important things.

Firstly, the city had to figure out water supply.

About 50% of the city's water supply comes from surface and groundwater, specifically the Verde River and the Salt River watersheds that are to the north and east of the city. And about 40% comes from the Central Arizona Project, which is a 541 km diversion canal that pulls water from the Colorado River and brings it through the state (terminating in Tucson). From what I have read, the CAP canal serves about 80% of the state's population.

Secondly, Phoenix needed A/C.

Air conditioning was first invented at the beginning of the 1900s, but it really didn't become ubiquitous until the second half of the century. And not surprisingly, it made the southwest of the US far more appealing. In 1950, Phoenix had just over 100,000 people. By 1960, the city had grown by about 311% to reach nearly 440,000 people. Apparently there was more construction in 1959 alone than from 1914 to 1945 combined. Air conditioning made Phoenix's summers bearable.

Air conditioning is such an interesting topic because it's one of those things that many of us take for granted. When you walk into a store or an office building in the summer, you expect it to be cool. But it hasn't really been around all that long and its invention has, in many ways, been instrumental in defining the modern city. Even something as common and banal as a glass office tower with large floor plates would not be possible/practical without air conditioning.

We can certainly debate how sustainable it is to urbanize and air condition desert climates like Phoenix, but there's no denying that air conditioning has had a profound impact on our urban landscapes. Diversion canals are pretty important too.

The postmortems surrounding Zillow's exit from the algorithmic home-flipping business are starting to surface. Here's an article from the WSJ and here's Matt Levine's take on it. The latter piece is very Levine-like and is called, "Zillow tried to make less money."

The obvious story is that Zillow's algorithms were not valuing homes correctly. But the story is more nuanced than this. In Q1 of this year, Zillow's home flipping business was actually more profitable than it had initially expected. And that's because its algorithms were consistently undervaluing homes. So when it did transact, it was doing so at favorable / low cost bases.

The problem was that the company was not transacting enough and there was a fear of losing ground to competitors like Opendoor. Apparently only about 10% of people who requested an offer from Zillow actually ended up accepting it. Margins were good, but volumes were too low.

So what Zillow did was tweak its algorithm to be more aggressive (see above chart from the WSJ). But this created the opposite problem: low/negative margins, higher volumes.

Once again, it shows you some of the challenges with bringing real estate online. The supply of homes is largely heterogenous and there are a lot of qualitative factors that play into what someone is willing to pay.

Things are happening in the algorithmic home-flipping business right now.

A few weeks ago I wrote about Zillow pausing this part of its business. It was then later revealed that the company was set to take a loss on many/most of the homes that it had purchased through this "iBuying" division. In October, it listed some 250 homes in Phoenix and on average they were priced about 6.2% below what they had bought them for.

So it is perhaps no surprise that today the company announced that it will be the exiting the business of buying high and selling low. Turns out this isn't good for business.

But does this mean that the model doesn't work or that Zillow simply didn't have its algorithms tuned correctly? Following the news, competitor Opendoor took to Twitter to reassure everyone that the digitization of real estate is still well underway:

Opendoor also announced today that it will be expanding technical hiring into Canada -- starting first with Toronto. The plan is to hire upwards of 100 people over the next several years. Presumably this is about access to talent, but presumably it also means that Opendoor is looking toward one day expanding into Canada.

Stay tuned.

Disclosure: I continue to be long $OPEN.