What's a good phrase for "refining a process which is hopelessly broken"?

I'm looking for a turn of phrase to describe a situation where the powers that be wish to continue making small improvements to a process which, due to deep-rooted flaws, will never be close to accurate.

I'm working on a process for estimating the states of business objects, and there's interest in fixing some edge cases which have been noticed. This could be worthwhile, but because of missing data we're already making heroic guesses. So if implemented, we might get from 80.0 to 80.1% accuracy. If we were at 99.5%, I'd be happy to get it to 99.6%, but in the current situation it doesn't seem productive.

We have some CS people on E.SE; what would you call it when someone wants to optimize a two-second process to load data, though once loaded it will be chewed on by an inefficient algorithm for hours?

Something like "bailing the ocean" or "fixing a leaky faucet in a burning building" is close to what I have in mind. There's also "rearranging deck chairs on the Titanic," but that carries an implication that the changes are wholly useless, and that events are coming to a head.

Solution 1:

We have some CS people on E.SE; what would you call it when someone wants to optimize a two-second process to load data, though once loaded it will be chewed on by an inefficient algorithm for hours?

In a CS context, I'd simply call this "premature optimization". The reason that phrase works is because, in that context, I'd expect most people to be familiar with this famous quote by Donald E. Knuth, or at least to have heard it referenced:

"Programmers waste enormous amounts of time thinking about, or worrying about, the speed of noncritical parts of their programs, and these attempts at efficiency actually have a strong negative impact when debugging and maintenance are considered. We should forget about small efficiencies, say about 97% of the time: premature optimization is the root of all evil. Yet we should not pass up our opportunities in that critical 3%."

If I wanted a more colorful term, I might call it something like "painting racing stripes on a tricycle" or, if it's about accuracy instead of speed optimization, perhaps "putting a laser sight on a nerf gun". A somewhat common security idiom would be "putting a vault door on a garden shed" or "a steel padlock on a cardboard box", etc.; the possible variations are endless.

Solution 2:

That's known as polishing a turd.

Solution 3:

You can say you're putting lipstick on a pig. Or, you can say you're tripping over a dollar to save a dime.