Steve Jobs claimed the best coders are 100x better than avg. Still true today? | Hacker News Reader

Steve Jobs claimed the best coders are 100x better than avg. Still true today?

6 pointswmilesn17 comments

Steve Jobs said in his 1995 'Lost' interviews that in general, "the range between average and the best is at most 2-1. In software, the difference between average and the best is 50-1 maybe 100-1."

Does this ratio still hold true today given the proliferation of equalizing tools such as jQuery (or even Javascript for that matter), Rails, Mongo, Chrome, Sublime, Codecademy, Stack Overflow, Coursera, [EC2, Heroku], etc.

View on HN

I'm currently working in a system that seems to have been written by neanderthals. Global variables everywhere, non-descriptive variable names, functions 500 LOC long, no main program loop, etc. It is written in Python, but my god, at this point its fucking Latin to me. Have spend two days trying to decipher it in order to bring an offline system back online (the system is what makes the company money). Realized its just easier to re-write this pile of shit into a program that's less than 500 LOC. Yes, from 5000+ LOC, to around 500. It will now be maintainable, and modular (and fucking readable). Tell me now, how much more is that worth?

PS. It was written by bad PHP code monkeys who lied about knowing Python (much less how to program).

Any chance of a looksie at the code? Off the record?

Doing so would mean that I value your curiosity more than I value my client, their business, their employees, my family, and the well being of everyone who depends on that system. So, no. And it's not cool to ask, either.

Seemed like a harmless enough request with an easy yes or no answer. I find him asking a lot cooler than your response. If the situation is so sensitive to everyones well being, you shouldn't even be discussing it on a public forum. Get over it.

It was an unfair request and one I made off the cuff.

With more time I would have either rephrased or not asked. I withdraw the request and apologise to both the above posters.

I don't particularly care about the sensitive portions of the poor 5000LOC but I am interested in how these things get so out of control.

I was some time ago trying to start a MSc in software engineering - focused on just these differences in productivity - and the biggest stumbling block is finding examples of bad code and coders to study. People who want to be part of a study tend to be in the upper quadrant and from my point of view boring.

Anyway, long story short, I posted before thinking

Take care

Why unfair? I have a friend who is an extremely competent programmer (and a very prolific writer) and he sends me - of his own initiative - snippets of code written by his cow-orkers (or at least he did, until he switched jobs and went into a company with much more respectable peers, so there's no fun to be had in that). You wouldn't believe what some people are capable of attempting to write. (Except, of course, you would, if you visit thedailywtf.com with any frequency.) Let's just say that, for example, an awful lot of people believe these days that they are competent enough to rewrite date-handling libraries in the Java ecosystem. Of course, the correctness of the result fails horribly.

But je is your friend and you don't ask him to drop the code on HN.

I really want to use the codebases of the bottom three quarter to help drag up the quality bar - but like I said it's a hard thing to arrange.

Added to this is that the lack of code literacy means these proprietary places can swim for years in the same circles never actually getting out from under.

If companies want competitive advantage from their technology they need to invest in software and code bases - reducing the pressure on delivery (CI or just cut back on the expectations) and focusing on getting better code. I mean no-one argues that a novel written by two hundred teenagers will be less intelligible and workable than something by Salinger. Why the hell would we expect different from code.

Not sure where I am going with this - other than writing good code is a big deal, and stuff that prevents it needs to get out of the way

In my opinion, it's even more true than it was 18 years ago.

The "equalizing" tools help the average programmer when everything works well, but they make it significantly more difficult for them when problems arise. Today's tools have given us many leaky abstractions.

There's a lot of peer reviewed software engineering research to suggest what Steve Jobs said is true, long before he said it.

http://skeptics.stackexchange.com/questions/7559/are-there-s...

I am sure Steve jobs wasn't referring to the quality of code produced. I would say he was talking about the results that where produced. So the tool you use might make it easier to work with. But if you are creating a highly optimized useless product it is a waste of time.

It depends in what) Some "coders" are just the best at pretending that they are the best LOL Proof: https://news.ycombinator.com/item?id=5066813

Things have changed a lot - more tools that allow bozos to get into engineering these days. There are just more wannabe engineers now so the % that is good is just smaller.

jQuery, Node, Sublime and Stack Overflow as equalisers?

They don't even narrow the gap by 1/10 of an inch.

My view is value as opposed to ability

Like a soccer player, some are x% better than others yet get paid 100x because that extra x% is able to win more games

Similarly a top developer may only be a bit better than you but that difference, when applied within a solid engineering company, can lead to business results 100x greater.

Measuring that is hard, getting it in a raise negotiation even harder.

Absolutely.

See Fact 2 from Facts and Fallacies of Software Engineering by Robert Glass.

There's a PDF online, but I suspect it's not legit.

100-1 might be an exaggeration, but the knowledge of different tools, methods, and patterns and how to apply them gives the experienced programmer the ability to produce code that does more, more quickly, and with fewer side effects than the inexperienced programmer. There is also a matter of natural talent: The best software engineers can design software that the average programmer would not be able to develop to the same quality if given years to work on it.

To give an anecdote, the place where I most recently worked had two developers who had been there since the beginning, a software engineer and another coder. The engineer on a bad day could quadruple the output of myself and the other guy in terms of both the number of changes and their value to the project.

Reply on news.ycombinator.com