Comment by paxy
3 days ago
Number of employees isn't a straightforward calculation.
- A lot of companies have part time, seasonal and contract workers (e.g. Amazon and Walmart). For some this even makes up the majority of their workforce.
- Some companies operate on a franchise model, and corporate employees are very limited. For example the 414K employees listed for Marriott aren't directly employed by them. But then you don't apply the same logic to McDonalds, where you are only counting the corporate employees.
- Holding companies (like Berkshire Hathaway) show the full employee count of their portfolio companies even through they may only own a small share. And again the logic isn't consistent because you are only counting 714 employees for Welltower, which is also a holding company.
Collecting incomplete or incorrect data and adding a hunded disclaimers at the end of the report is standard LLM behavior. Actually valuable analysis of this sort needs a lot more human effort.
Fair point, I was aware of that and mention it in the accompanying blog https://bigwhalehunting.com/blog/workforce/
Towards the bottom:
> Employee counts come from company reports via Yahoo Finance, with Wikipedia used where figures weren't available. They're not necessarily from the same date, and companies don't all count employees in exactly the same way.
> As the McDonald's and Welltower examples show, a company's reported workforce doesn't necessarily represent everyone whose labour contributes to its business.
Well, alright. I'm just not sure how useful this is. There's no way to normalize them but the data's presented as though they were. You also sum employees across sectors which compounds the errors.
> surprised me
> I was aware of that and mention it
I'm confused.
I imagine Tesla has a vast amount of contract workers with all the construction they do.
The blog post OP linked puts more context into it, and I think it also makes it clear that they’re not trying to make an apples to apples comparison. It’s a tool that can aid in visualizing how different companies can be structured.
Except this visualization is literally an apples to apples comparison
Half of Google's employees are TVCs (temps, vendors, contractors)
Berkshire reports employees across its consolidated subsidiaries, while Welltower's 335,000+ operating-partner employees work for separate companies. I'm using reported employee counts consistently, but you're right that different corporate structures make comparisons imperfect.
> Collecting incomplete or incorrect data and adding a hunded disclaimers at the end of the report is standard LLM behavior. Actually valuable analysis of this sort needs a lot more human effort.
Or it's classic modern journalism that gets clicks. LLMs are just mimic'n human behavior.
No one is calling this journalism especially not me.