To be less glib, I think the article makes it clear just how inordinately difficult blockchains make their lives when they have to build, from scratch, a reliable definition of time.
If you could have a trusted time source, you could at a stroke get rid of one of the most egregious flaws in bitcoin, its vast wastage of energy, because of its 'proof of work' mining. Bitcoin miners are running at full pelt 24/7, picking random numbers and doing math on them in the hope of striking lucky and discovering the next block. The mining difficult is artificially picked so that blocks are discovered about every ten minutes.
However, imagine if bitcoin miners could all declare a cease-fire, for 9m59s of those ten minutes, i.e. they just do nothing for that long, consuming virtually no energy, and then they all furiously mine at full speed for the last second (with the difficulty adjusted so the global mining should take a second). This process would be just as fair as the current one; if you have more/faster hardware, you will still mine more, in the same proportion as before. The total energy usage would be reduced 600-fold!
But - the reason you can't do this is that there's no way of stopping cheating. Who can tell if the other miners are really idling for those 9:59? This is where the time source comes in. Imagine you have a trusted time source: it could, every ten minutes, broadcast a random number. Miners would have to listen for this number, then mine a block containing it, to prove that they didn't start work early. Problem solved!
Back to the real world: just about everyone can agree on time; we have NTP and can even use certificates to authenticate clock sources, we can even use multiple sources to make it harder for time to be faked. So why can't blockchains do the same thing? Why don't they use multiple time oracles to stop the colossal energy wastage?
And, when bitcoiners tell you that relying on a centralised source, or even a quorum of sources, is completely unacceptable, why then are the same people happy to use smart contracts where the use of centralised oracles is apparently both acceptable and commonplace?