Finding a bitcoin block is like picking a number out of a hat. A hash is like reaching in the hat and picking out a single number. All miners on the network are reaching in the hat over and over, as fast as they can, hoping to pick out the right number.

When a miner picks out the right number, observers of the network, like us, can only see that the miner holds the right number. We cannot see how many reaches the miner made into the hat in order to find the right number. Sometimes miners get luckier than expected, pulling the right number out of the hat after only a few reaches. Sometimes miners make as many reaches as we'd expect (or more) before pulling out the right number. It could've taken the miner one reach, two, or far more than a trillion trillion reaches. We, the network, only know that the miner holds the right number and how long it took them to pull it out of the hat.

The bitcoin network is designed so that it takes miners, on average, 10 minutes to pull the right number out of a hat after a new hat is introduced. If for most recent hats miners pulled the right numbers out faster than expected, that probably means bitcoin miners made more reaches into the hat per second than expected. That is, miners probably increased their reachrate.

We say probably increased their reachrate, because it's possible miners didn't increase their reachrate at all. Miners might just be on a lucky streak. Lucky streaks don't last forever though, and the longer a lucky streak lasts, the more likely it is that miners have really increased their reachrate.

So we can never with 100% confidence say what the hashrate, or reachrate in our metaphor, is. We can only observe how fast miners are finding blocks and make informed guesses about the network's hashrate based on that. The longer we observe miners doing anything though, the better our guesses about miners get. The longer we observe miners finding blocks faster or slower, the more confident we can be that the hashrate has probably increased or decreased.