> While YouTube, or at least YouTube Shorts, still pays creators more...
I would imagine this matters a lot to the largest creators which would be the ones that likely have the most upfront and reoccurring costs (in order to make quality, long form videos).
The 30 minute podcast boils down to the concept of a rent ratio (cost to buy house / house to rent house for a year).
> I was interviewing Mark Zandi, who’s an economist who’s been looking at housing for many, many years. And he said, right now, with current interest rates, he thinks the break even point is about 18. If you can find a rent ratio of 18 or lower, it will often make sense to buy.
> Now, spoiler alert, you’re not going to find a rent ratio of 18 in just about any housing market in the country because housing is so expensive right now. And that’s part of the reason why, for so many people, it really does make sense to rent right now.
Essentially, buy if { cost of buy }/(18*12) < { cost to rent/month }, otherwise rent.
> According to data from the National Association of Realtors, the median price for an existing home — one that's already standing, not new construction — was $387,600 as of November 2023. - bankrate
For the average US home price of $387,600, the recommendation would be to buy if the rent is more than $1794/month.
Fig. 5. shows 10 participants using ChatGPT with 40% (4/10) of responses being judged as "Expert" and 15 participants not using ChatGPT with 53% (8/15) of responses being judged as "Expert"...
The data simply isn't significant and the design of the study has a very subjective measure of what it considers to be an expert response.
The article is 6 years old but the CSV Injection still works on Google Sheets and Excel. (I imported the example csv file into both platforms and the equations were executed)
Great reminder to be vigilant. Pride cometh before a fall:
> Are they a technically savvy user? Then it is even worse. They know the CSV format is just text data, there can’t possibly be anything harmful in there. Guaranteed.
> Roughly a month into the trip, the team captured a sonar image of the plane-shaped object about 100 miles from Howland Island — but didn't discover the image in the submersible's data until the 90th day of the voyage, making it impractical to turn back to get a closer look.
To be a skeptic, this sounds like confirmation bias.
Regardless, ChatGPT likes to give me recommendations and mentions by city by name (even though it is sometimes wrong and thinks I'm ~300 miles away).
I like the culprit is 4o using web search behind the scenes even when not explicitly enabled.
reply