为什么当我添加 thread.join（） 时，这个多线程代码的行为更安全？

问：

import threading

# Shared variable
shared_variable = 0
NUM_THREADS = 999
NUM_INCREMENT = 1_000_000


# Function to increment the shared variable
def increment():
    global shared_variable
    for _ in range(NUM_INCREMENT):
        shared_variable += 1


# Creating multiple threads to increment the shared variable concurrently
threads = []
for _ in range(NUM_THREADS):
    thread = threading.Thread(target=increment)
    threads.append(thread)

for thread in threads:
    thread.start()

for thread in threads:
    thread.join()

# Display the value of the shared variable after concurrent increments
print(
    f"The value of the shared variable is: {shared_variable}, expected : {NUM_THREADS * NUM_INCREMENT}"
)

当我加入线程时，此代码始终打印预期的数字。但是，如果我注释此代码，则某些增量会失败（这是我实际期望的结果！

# for thread in threads:
#    thread.join()

为什么当我添加连接时，此代码以线程安全的方式运行？