Agreed. I definitely thought that at first, thinking some of them seemed very off. Glad I read these comments. It’s especially confusing considering where some things are in the map that it seams almost believable for example that NY/NJ are made up mostly of mostly urban and commercial areas.

But it is a good chart (not map) for what it’s intended to show with some perspective provided in proper labeling.

Agreed. The concept of judging vehicle quality by number of recalls is severely flawed for this very reason. My Subaru Impreza has had a number of recalls for a variety of trivial things, but I’ve had only one actual issue with it in 65k miles and have spent relatively little on maintenance. Comparing that to the Audi A4 I had before this car which required maybe one recall in similar mileage but I was constantly fixing major items from leaks, broken drive related components, etc.

Neither had any motor related issues so far, aside from burning oil in the Audi. But by number of recalls? That Audi was great! But they also had a number of lawsuits filed in attempt to get them to actually recall the multitude of problems. The one that it actually had was the result of them losing such a suit, but so many years later it really didn’t matter.

So yeah, terrible metric to track. At this point, I’d rather see that the company has a dozen recalls on their vehicles than zero.

Edit: I should clarify. That being said, I do believe Toyota actually makes a solid car the first time. Boring, but quality is a huge focus for them. I’m still hesitant to trust recall counts though and I don’t think I’d trust Mercedes number as a valid quality metric.

I use nginx to handle the routing to different systems and docker ports so it helps when I fiddle with that config to see “server online” that I shoved in a basic html page for the base domain.

I used to run Nextcloud there but I’ve had so much trouble with nextcloud that I barely use it at all now and I worried it was too easily discoverable for something I wasn’t confident was configured properly.

I’m interested to hear more. Cherry picking of data for which drugs to pursue, I had always assumed was focused on trying to push the most viable drug through to see if the benefits could outweigh the risks, costs, etc. For example, many cancer drugs come with insane side effects but they also offered the best benefits. The side effects could have disqualified the drug early in the process, but it was still worthwhile to see that there is something there that helps. My understanding is it goes like this: we found a drug that can cure cancer 40% of the time, but there’s a 50% risk of liver failure. These are fake numbers, but they likely know most of this before sending it through clinical trials, but they also know that someone literally in agony and about to die from this particular cancer would be glad to take those odds. From there, it should open up the door to more focus on the positive results as more are cured from the disease even if it causes liver failure in the select few that were about to die and took the risk.

I’m not sure I follow your conclusion that there’s more incentive to almost pass than to have nothing? That’s generally the expectation everywhere: show you’re doing something. What leads you to say this is different in biotech and what is it harming?

The fudging numbers part is really the most alarming, but that would only go so far. The FDAs needs to approve it, and I was under the impression that’s done independently. So if a drug company wants to waste money trialing things that won’t get approved, that’s just a loss of money to run the program and trials and would have no net gain unless they’re looking to uncover problems they can work out on a new drug.

I don’t work in biotech, so I’m hoping you can elaborate a bit more if my trust is misplaced.