[HarryHirsch]

Perhaps related: https://orgprepdaily.wordpress.com/2007/04/22/on-keeping-you...

This is a chemist's view from the trenches. A manager who hasn't been brought up in the engineering mindset wouldn't even be able to understand any of the points in that blogpost, which partially explains the trouble that pharmaceutical companies are finding themselves in. You cannot bribe or cheat observable reality.

[lutorm]

As an engineer, I sort of disagree with this. Doing a shoddy job and trying to get away with it is inexcusable, but "good enough" is the gold standard in engineering. No one is helped by spending development time and money by making things better than they have to be.

Sometimes you have to go with a less capable result because the alternative will take too long or cost too much to develop, or (since we're talking aerospace here) weigh too much.

The trick, of course, is to know exactly how good is enough.

[SilasX]

Right, there's a difference between, "because of these unavoidable constraints, we are deliberately making this design compromise" vs "meh, no one feels like following the standards so we'll just pressure everyone into pretending they don't exist".

[nate_meurer]

> "good enough" is the gold standard in engineering.

This doesn't really make sense given any common meaning of good enough. A well engineered system or product is optimized. If you want to say that good enough means solving the inherent tradeoffs between safety margins and cost with careful design and precise specs, then you're not talking about the same thing as that blog post, nor would that accurately describe the way Boeing appears to design commercial aircraft nowadays.

[lutorm]

Isn't the common meaning of "good enough" "sufficiently good to meet my requirements, even though I can imagine it being even better"?

But you're right, I'm clearly not talking about the same thing as that blog post. What I'm arguing is that there's this appropriation of language where people use "good enough" to mean exactly "not good enough", and I don't think that's helpful.

[TylerE]

Development time and costs are major dimensions to be optimized.

[karmelapple]

Fully agreed. And “good enough” for some software that does not interface with human digestive systems is a whole lot different than pills that people might take daily.

[onceUponADime]

There are some companys who manage it though, by basically getting bi-furicated intwo two seperat internal divisions competing against one another.

Basically if you have a safe seat- or monopoly, you must learn to play catch against yourself or else.. boing, and important parts go flying.

[pkaye]

> There are some companys who manage it though, by basically getting bi-furicated intwo two seperat internal divisions competing against one another.

Sometimes that competing becomes backstabbing between the divisions.

[sct202]

And then the company falls apart when you get a real competitor and the divisions are too busy attacking each other.

[thombat]

And Nokia was one such company, in the latter days of S60 smartphones. One (of many) re-orgs produced three main phone divisions called something like: entertainment, executive, and enterprise, with high-level features partitioned between them [1]. So "entertainment" gets the high-quality auto-focus camera because that's great for blogging hipsters, and "executive" gets the barcode reading software so suits can scan the QR codes on business cards [2] ... except they couldn't because without a decent auto-focus camera the QR code needed to be printed on A3 paper and stuck on the wall to be scanable. Yes, this was discovered in internal testing. Yes, it shipped anyway.

Down in the trenches we made bitter jokes about "three bald men fighting over a comb" as we watched our Lords and Masters engage with the looming competition by making snide public remarks about "that fruit company", having long before killed the touchscreen prototypes. But at least Boeing would have handled this situation differently: we'd have been fired for those jokes.

[1] and what if you're an "executive" working in an "enterprise" who likes to kick back with some "entertainment" on the commute? KA-CHING! you're gonna buy three Nokia phones! Or zero, more likely.

[2] that was seriously advanced as a leading use case. Quite probably by an in-house futurist.

[jessaustin]

This is a great idea. Can we think of any examples where this has worked? Famously, it didn't work at Kodak when they invented digital photography.

[mark-r]

Kodak gets picked on unfairly I think. When you look at the revenue streams before digital photography took off, you can see that Kodak had their fingers in every part of it - cameras, film, processing, printing. They made money every time you took a picture. There's simply no revenue stream in digital that's big enough to replace that. It's impossible to shrink a company by 90% gracefully.

[jessaustin]

Kodak may have been boned no matter what, but it's still not the example I seek, of a firm that successfully "played catch against itself".

[mark-r]

I was just pointing out that there were reasons it wasn't successful beyond just not being a viable approach.

[ohyes]

I guess they could have just invented Instagram or something

[mark-r]

That would have been interesting, but they still would have needed to figure out how to monetize it.

[manicdee]

There is a clear disparity in the understanding of the phrase “good enough.”

The “good enough” from an engineering perspective is, “meets the customer’s specifications and is the best we can do given the constraints of time and money.” It’s about the quality of the product.

The “good enough” in this essay relates to the expenditure of emotional effort in getting your work done to a known standard while management are applying pressure for you to reduce the amount of time and money invested. “Good enough” in this instance relates to the effort exerted. At some point you are not going to try any harder because further effort is counterproductive to your career prospects or financial well-being.

This “good enough” is the lab full of technicians who know the product doesn’t even come close to the specifications it is being marketed towards, but they continue to work because they need food on the table. It’s “good enough” as a dictum from management, not as an informed decision from the product team.

[HarryHirsch]

No, the "good enough" in Milkshake's essay means not conforming to accepted standards, it's good enough only for management, who consciously or unconsciously intend to deceive a client downstream and sell up before someone else notices.

Elsewhere, Milkshake has a story about a biological chemist on a project to develop a delivery vehicle for cancer drugs. This fellow found his results unconvincing, continued working on the problem and ended up demonstrating convincingly that in vivo the vehicle did not behave as intended. Now the company had a problem (the FDA demands that relevant results be reported) because management intended to sell the IP to a hapless bidder, and the chemist's action pierced the veil of plausible deniability: https://orgprepdaily.wordpress.com/2017/08/24/breaking-bad-i...

[manicdee]

Would you say this researchers efforts were counterproductive to his career goals?

[redis_mlc]

> but "good enough" is the gold standard in engineering

Your post does a dis-service to the 350+ who died.

Nobody agrees that using one external sensor on the 737 MAX that is subject to weather, bird strikes and ground handling damage was nearly good enough.

[Spivak]

You're arguing from two different perspectives. Engineering is about producing an deliverables that just meet spec. If someone orders a temperature sensor that they ask to be 1° accurate then you are wasting your time designing something better^.

However the engineers in question at Boeing clearly failed to produce a design that performed to spec in normal operating conditions.

^ Yes, there are scaling factors where it might be cheaper to design a 0.1° accurate sensor and sell it to everyone because it would be more effort having two production lines but that's an optimization.

[redis_mlc]

> You're arguing from two different perspectives.

Your comment doesn't make any sense in the English language.

Nobody ordered a sensor system that was unreliable, so no idea what you're talking about.

[lutorm]

Exactly. It was obviously not "good enough", so they failed to meet that standard. That's my point.