That rather makes sense when the bit of kit isn't something you can just send a man out in a van to repair !
They can replace repeaters, but they have to go out, drop a (high tech) grappling hook to fish up the cable, cutting the cable and dropping one end back on the seabed in the process. Then they can reel in the other end till they reach the broken device. They can replace the device, splice in a new longer bit of cable, go back and fish for the end they left behind, join the ends up, and finally drop a big loop back onto the seabed (making it easier for wayward ships to get it with their anchors).
The cable is now longer, and has been completely out of service while they are doing it. Yes, I think it's a good idea to do everything you can to make the stuff reliable and tolerate faults.
Incidentally, the very first transatlantic cable did get to send a few messages before it failed - including (IIRC) one cancelling the return of troops from Canada to the UK as they were needed elsewhere (and thus saving weeks of voyage back here only to be sent back out again). But the message rate was measured in symbols/minute - so something like one character per minute maximum. At the time, there was no understanding of transmission lines - and the assumption was they they just needed to "use more power". As it was, they were using something like a kV at one end, and getting a barely visible deflection on a galvanometer at the other.
But when they turned the voltage up too far, they had an insulation breakdown, and no method back then to locate the fault or repair it.
One theory is that the initial laying attempt was abandoned, and most of the cable was laid up on a dockside for a year before they tried again. In the heat of the sun, the gutta percha insulation may have softened, allowing the centre conductor to move so it was no longer central - thus reducing the effective thickness of the insulation.
Not long afterwards, IIRC it was
Heaviside who came up with the theory (and equations) relating to transmission lines - this allowing the speed to increase to symbols/second.
Interestingly, at a talk I went to on this not long ago, the presenter put up a slide showing a history of sub-sea cables and the companies that put them in - along with a bit of "and X was absorbed into Y, and Y was taken over by Z, ...". That first cable can be traced through to what was until recently called Cable and Wireless (though those of us with the "experience" of dealing with them prefer Clueless and Witless) - they are now part of Vodamoan (oops, Vodafone).