I'm not sure about the time scale you're referring to, but I have some expeirence with dog training and I've been interested in dog training history lately, so maybe I have insight for you. Also, I want to qualify this whole tirade by saying this is a USA-centric breakdown; other countries have different cultural histories with their dogs, and while the underlying animal behavior is the same, I can't speak to whether dogs in other countries are "well" or "poorly" trained.
Prior to the 1900s, dogs weren't really thought of as companion animals the way they are now. Dogs were usually from working lines-- hunting dogs, setters, pointers, terriers, ratters, herders, shepherds, guard dogs, sled dogs, etc. They were considered somewhat adjacent to livestock. In these situations, dogs were often "trained" by their breeding. You don't have to tell a working line rat terrier to kill rats, they just do. Sheepdogs will herd children if there aren't sheep around. Just try keeping a working line husky from pulling in a harness... you can do it, but it's working against it's nature. Mostly around this time, a person had multiple dogs of breeds with natural instincts to do the job they wanted them to do, and the dogs did it. The ones that did it best were bred by their owners, and the next generation was better than the last. It's also important to note that the major written documents describing dog training at this time mostly emphasized rewarding the dogs with meat and praise when they are good, and ignoring them when they are bad.
During and around WWII, there was a new interest in training dogs for policing, warfare, and personal protection. It became more common to have one-dog-one-handler arrangements, and since most working lines of guard dogs were more "bark at intruders and bite strangers" kinds of dogs instead of "dutifully and silently stand by until ordered to kill" dogs, there was an interest in developing training methods to achieve the desired result without needing to breed new working lines.
From this desire during WWII, two schools of thought arose. One was the "traditional" method (not very traditional after all...) which arose from trainers like William Koelher. These methods emphasized discipline, "corrections", and punishment. The other school of thought had its roots from behaviorists like Marian Breland Bailey (an advisee of BF Skinner) that illustrated the power of operant conditioning and positive reinforcement. They both started around the same time (1930s-1960s) but for one reason or another the traditional methods were more popular, and the reinforcement methods were seen as lesser "tid-bit training techniques" based in "the prattle of 'dog psychologists'".
It turns out they were both working with a similar framework-- dogs learn by associating an action or stimulus with a positive or negative outcome. The argument was whether positive or negative outcomes were better at inducing learning gains. At this point, mountains of research shows that positive reinforcement wins out every time, meaning that the behaviorists were more correct than the traditionalists.
Still, as I mentioned, the traditional methods were more popular for a long time. People still think they need to "be an alpha" or leader to their dogs, that they need to discipline the dog so it respects them, that punishing the dog is the way to achieve good behavior. Choke and shock collars, leash corrections, and "alpha rolls" are still common training techniques despite the evidence that they are counterproductive. Additionally, you'll remember what I said about the behaviorist/reinforcement methods being more aligned with training techniques recorded before WWII-- when farmers were training herding dogs, they weren't "alpha roll"ing them, they were giving them meat when they did their job and ignoring them when they didn't.
Anyway that's a whole fucken essay in itself, but the point I'm trying to make is this: prior to WWII, dogs were trained by being paid in daily food and by having the chance to breed. Many working dogs are still trained like this, perhaps giving you the impression that dogs "used to be trained well". Companion dogs are a more modern development and there continues to be two schools of thought about how to train them. People who look deeply into evidence-based dog training methods train their dogs with positive reinforcement-- these dogs are usually what we consider "well trained" dogs, and overwhelmingly these dogs exist in affluent areas where dog owners have the money to pay for expensive trainers, and where they have the free time to train the dog consistently. As class disparity grows, it is becoming more common for people in poorer areas to lack access to the education about the best methods, so they tend to default to "traditional" methods that were more popular in the 20th century. These dogs are... less "well trained". Even if someone wants to put in a lot of effort to learn how to train dogs, they might just not have access to the most up to date knowledge. Additionally, there's evidence that dogs trained with these methods are more susceptible to a lack of generalization than reinforcement trained dogs, which is to say they might act fine in most situations, but they act worse (more fearfully, less predictably) in novel scenarios. That's part of why you might see "well trained" dogs who suddenly and disasterously act out.
One last side note: often dogs who are reactive (the term for dogs who freak out and start screaming when they see a person or a dog or a bike, etc.) are not necessarily untrained. Reactivity is a fear response; you can imagine they might be like a normal human with a spider phobia. They might be 100% perfectly behaved in every situation... except for when a dog walks by. In this situation, the other dog is like a spider.
Traditional training might suggest that you try to order the dog to stop freaking out and punish them if they don't stop when they see another dog, but that's like punishing someone with a spider phobia for freaking out when they see a spider. The reinforcement methods instead try and convince the dog that other dogs (spiders) are actually harmless. This is shown to reduce reactivity much more than punishment. Still, reducing reactivity is like really really hard, just as fixing phobias in humans is. Even if someone is working very hard with training and using the best available techniques, the dog might still freak out when they see another dog (thus looking like they "aren't trained", according to your post).
And LAST last note, maybe the difference you're perceiving is from covid? A lot of people got a lot of dogs but couldn't take them out to socialize and train them due to lockdown. Additionally, during covid a lot of adoption agencies literally ran out of dogs, meaning that dogs that would usually be euthanized because of behavioral issues were instead adopted out to families. Compounded with a lack of socialization, and the fact that many people still use "traditional" training methods, maybe you're just seeing a lot of reactive, fearful dogs? Hopefully that will improve over time!
Anyway thanks for reading my whole fucken essay, lol... I wrote this while on a plane so I guess that's why I was bored enough to write this much. Hope you get something out of it!