A lot! Think of the legal system like a CPU. The legislature is a big, slow HDD. The courts are like DRAM, much faster than the legislature but still slower than ... The USPTO. The USPTO run by the Director is like L3 cache in this analogy - very quick and close to the processing. Maybe that makes the examiners L1/L2 cache. The point is, when the L1/L2 cache has a cache miss, it goes to the L3, right? That happen all the time and is pretty quick, it's a tight feedback loop. Reaching out to DRAM is much slower and less frequent, and same goes even more for mass storage. In this analogy, the instructions are the laws and the processing is everything the USPTO does.
The analogy breaks down a little bit from there, because the laws are not perfectly communicated between each layer, it's like someone inserted a fuzzer in the bus at each stage so that getting from the HDD to the CPU is both slow and unreliable. There's only one step of fuzzing between the L3 cache (the director) and the CPU - much clearer, more direct, quicker, etc.
Beyond the somewhat silly analogy, our legal system is set up to grant quite a bit of deference to administrative agencies like the USPTO, referred to as Chevron deference if you want to look it up. The judicial and legislative branches do provide feedback through lawmaking and court rulings, but each stage of the feedback loop is indirect and slow. The courts don't really tell the USPTO what to do (in general), they just decide specific cases presented to them. Then, the USPTO has to figure out how to act in accordance with those rulings, which isn't always clear. So then the Director has to figure out what the courts really meant and then implement some new internal procedures and rules to adhere to those rulings, which is a bit like herding cats. And they have to fill in a lot of gaps too, because again it's not like the courts sit down and figure it all out, they just rule on the narrow issues presented to them.
Let's end with an example. What is eligible to be patented? The legislature last weighed in on this question in 1952 with 35 U.S. Code ยง 101. The Supreme Court hears a case on this topic like ever 5 years or so on average (roughly). The last big one was Alice in 2014. Then, the lower courts (Fed. Cir.) started to get a trickle of cases about what that Supreme Court ruling really meant over the years, and the Fed. Cir. hears these somewhat frequently, like every couple of months or so. But, the whole time, the USPTO has been trying to follow the ruling of Alice from 2014, and it's been a moving target to some extent based on Fed. Cir. rulings. Go look at the USPTO's 101 training page [1]. This is what the Director and those under her have been doing for the last few years. Scrambling to figure out what the courts are saying and devising training and rules to implement it. There is a lot of leeway and gap-filling in this process! And ultimately, if you want to get a patent issued, what the Director says is what matters. What happens after that is all in the judicial branch, but patents are presumed to be valid and the Director is the one who issues patents. Her job is incredibly important to the functioning of the US patent system, and doing it well is important to all of us. For the most part, Director Lee was respected and admired by the patent bar and she will be missed!
EDIT: just for anyone who reads my conclusion and then goes to the IP Watchdog link and thinks I'm off base, just be aware that that site and it's author are somewhat extremist in their views and don't necessarily represent the majority opinion of all patent practitioners. My opinion is that the PTAB and its implementation has been an overall win for patents and the country at large. As is the crack down on 101. The end result of both is that prosecutors are putting much more effort into writing 'good' patents to navigate these treacherous waters and that can only be a good thing. And I've also seen lots of bad patents killed by the PTAB, which again is only a good thing.
The analogy breaks down a little bit from there, because the laws are not perfectly communicated between each layer, it's like someone inserted a fuzzer in the bus at each stage so that getting from the HDD to the CPU is both slow and unreliable. There's only one step of fuzzing between the L3 cache (the director) and the CPU - much clearer, more direct, quicker, etc.
Beyond the somewhat silly analogy, our legal system is set up to grant quite a bit of deference to administrative agencies like the USPTO, referred to as Chevron deference if you want to look it up. The judicial and legislative branches do provide feedback through lawmaking and court rulings, but each stage of the feedback loop is indirect and slow. The courts don't really tell the USPTO what to do (in general), they just decide specific cases presented to them. Then, the USPTO has to figure out how to act in accordance with those rulings, which isn't always clear. So then the Director has to figure out what the courts really meant and then implement some new internal procedures and rules to adhere to those rulings, which is a bit like herding cats. And they have to fill in a lot of gaps too, because again it's not like the courts sit down and figure it all out, they just rule on the narrow issues presented to them.
Let's end with an example. What is eligible to be patented? The legislature last weighed in on this question in 1952 with 35 U.S. Code ยง 101. The Supreme Court hears a case on this topic like ever 5 years or so on average (roughly). The last big one was Alice in 2014. Then, the lower courts (Fed. Cir.) started to get a trickle of cases about what that Supreme Court ruling really meant over the years, and the Fed. Cir. hears these somewhat frequently, like every couple of months or so. But, the whole time, the USPTO has been trying to follow the ruling of Alice from 2014, and it's been a moving target to some extent based on Fed. Cir. rulings. Go look at the USPTO's 101 training page [1]. This is what the Director and those under her have been doing for the last few years. Scrambling to figure out what the courts are saying and devising training and rules to implement it. There is a lot of leeway and gap-filling in this process! And ultimately, if you want to get a patent issued, what the Director says is what matters. What happens after that is all in the judicial branch, but patents are presumed to be valid and the Director is the one who issues patents. Her job is incredibly important to the functioning of the US patent system, and doing it well is important to all of us. For the most part, Director Lee was respected and admired by the patent bar and she will be missed!
[1] https://www.uspto.gov/patent/laws-and-regulations/examinatio...
EDIT: just for anyone who reads my conclusion and then goes to the IP Watchdog link and thinks I'm off base, just be aware that that site and it's author are somewhat extremist in their views and don't necessarily represent the majority opinion of all patent practitioners. My opinion is that the PTAB and its implementation has been an overall win for patents and the country at large. As is the crack down on 101. The end result of both is that prosecutors are putting much more effort into writing 'good' patents to navigate these treacherous waters and that can only be a good thing. And I've also seen lots of bad patents killed by the PTAB, which again is only a good thing.