Slashdot Mirror
New Book Describes 'Bluffing' Programmers in Silicon Valley (theguardian.com)
Long-time Slashdot reader Martin S. pointed us to this an excerpt from the new book Live Work Work Work Die: A Journey into the Savage Heart of Silicon Valley by Portland-based investigator reporter Corey Pein.
The author shares what he realized at a job recruitment fair seeking Java Legends, Python Badasses, Hadoop Heroes, "and other gratingly childish classifications describing various programming specialities." I wasn't the only one bluffing my way through the tech scene. Everyone was doing it, even the much-sought-after engineering talent. I was struck by how many developers were, like myself, not really programmers, but rather this, that and the other. A great number of tech ninjas were not exactly black belts when it came to the actual onerous work of computer programming. So many of the complex, discrete tasks involved in the creation of a website or an app had been automated that it was no longer necessary to possess knowledge of software mechanics. The coder's work was rarely a craft. The apps ran on an assembly line, built with "open-source", off-the-shelf components. The most important computer commands for the ninja to master were copy and paste...
[M]any programmers who had "made it" in Silicon Valley were scrambling to promote themselves from coder to "founder". There wasn't necessarily more money to be had running a startup, and the increase in status was marginal unless one's startup attracted major investment and the right kind of press coverage. It's because the programmers knew that their own ladder to prosperity was on fire and disintegrating fast. They knew that well-paid programming jobs would also soon turn to smoke and ash, as the proliferation of learn-to-code courses around the world lowered the market value of their skills, and as advances in artificial intelligence allowed for computers to take over more of the mundane work of producing software. The programmers also knew that the fastest way to win that promotion to founder was to find some new domain that hadn't yet been automated. Every tech industry campaign designed to spur investment in the Next Big Thing -- at that time, it was the "sharing economy" -- concealed a larger programme for the transformation of society, always in a direction that favoured the investor and executive classes.
"I wasn't just changing careers and jumping on the 'learn to code' bandwagon," he writes at one point. "I was being steadily indoctrinated in a specious ideology."
The author shares what he realized at a job recruitment fair seeking Java Legends, Python Badasses, Hadoop Heroes, "and other gratingly childish classifications describing various programming specialities." I wasn't the only one bluffing my way through the tech scene. Everyone was doing it, even the much-sought-after engineering talent. I was struck by how many developers were, like myself, not really programmers, but rather this, that and the other. A great number of tech ninjas were not exactly black belts when it came to the actual onerous work of computer programming. So many of the complex, discrete tasks involved in the creation of a website or an app had been automated that it was no longer necessary to possess knowledge of software mechanics. The coder's work was rarely a craft. The apps ran on an assembly line, built with "open-source", off-the-shelf components. The most important computer commands for the ninja to master were copy and paste...
[M]any programmers who had "made it" in Silicon Valley were scrambling to promote themselves from coder to "founder". There wasn't necessarily more money to be had running a startup, and the increase in status was marginal unless one's startup attracted major investment and the right kind of press coverage. It's because the programmers knew that their own ladder to prosperity was on fire and disintegrating fast. They knew that well-paid programming jobs would also soon turn to smoke and ash, as the proliferation of learn-to-code courses around the world lowered the market value of their skills, and as advances in artificial intelligence allowed for computers to take over more of the mundane work of producing software. The programmers also knew that the fastest way to win that promotion to founder was to find some new domain that hadn't yet been automated. Every tech industry campaign designed to spur investment in the Next Big Thing -- at that time, it was the "sharing economy" -- concealed a larger programme for the transformation of society, always in a direction that favoured the investor and executive classes.
"I wasn't just changing careers and jumping on the 'learn to code' bandwagon," he writes at one point. "I was being steadily indoctrinated in a specious ideology."
Older generations called this kind of fraud "fake it 'til you make it."
The moment you said "contractors", and you have lost any sane developer. Keep swimming, its not a fish.
Ah yes, the good old 80:20 rule, except it's recursive for programmers.
80% are shit, so you fire them.
Soon you realize that 80% of the remaining 20% are also shit, so you fire them too.
Eventually you realize that 80% of the 4% remaining after sacking the 80% of the 20% are also shit, so you fire them!
...
The cycle repeats until there's just one programmer left: the person telling the joke.
---
tl;dr: All programmers suck. Just ask them to review their own code from more than 3 years ago: they'll tell you that it sucks, and the person who wrote it should be fired.
I have a theory that
10% of people are good at what they do. It doesn't really matter what they do, they will still be good at it, because of their nature. These are the people who invent new things, who fix things that others didn't even see as broken and who automate routine tasks or simply question and erase tasks that are not necessary. If you have a software team that contain 5 of these, you can easily beat a team of 100 average people, not only in cost but also in schedule, quality and features. In theory they are worth 20 times more than average employees, but in practise they are usually paid the same amount of money with few exceptions.
80% of people are the average. They can follow instructions and they can get the work done, but they don't see that something is broken and needs fixing if it works the way it has always worked. While it might seem so, these people are not worthless. There are a lot of tasks that these people are happily doing which the 10% don't want to do. E.g. simple maintenance work, implementing simple features, automating test cases etc. But if you let the top 10% lead the project, you most likely won't be needed that much of these people. Most work done by these people is caused by themselves, by writing bad software due to lack of good leader.
10% are just causing damage. I'm not talking about terrorists and criminals. I have seen software developers who have tried (their best?), but still end up causing just damage to the code that someone else needs to fix, costing much more than their own wasted time. You really must use code reviews if you don't know your team members, to find these people early.
to find these people early
and promote them to management where they belong.
If I'm a plumber, and I don't know anything about the engineering behind the construction of PVC pipe, I can still be a good plumber. If I'm an electrician, and I don't understand the role of a blast furnace in the making of the metal components, I can still be a good electrician.
The analogy fits. If I'm a programmer, and I don't know how to make an LZW compression library, I can still be a good programmer. It's a matter of layers. These days, we specialize. You've got your low-level programmers that make the components, the high level programmers that put together the components, the graphics guys who do HTML/CSS, and the SQL programmers that just know about databases. Every person has their specialty. It's no longer necessary to be a low-level programmer, or jack-of-all-trades, to be "good."
If I don't know the layout of the IP header, I can still write quality networking software, and if I know XSLT, I can still do cool stuff with XML, even if I don't know how to write a good parser.
Then you get someone good at sorting arrays while picking out prime numbers, but potentially not much else.
The point of the test is not to identify the perfect candidate, but to filter out the clearly incompetent. If you can't sort an array and write a function to identify a prime number, I certainly would not hire you. Passing the test doesn't get you a job, but it may get you an interview ... where there will be other tests.
That seems about right to me.
I have a lot of weaknesses. My people skills suck, I'm scrawny, I'm arrogant. I'm also generally known as a really good programmer and people ask me how/why I'm so much better at my job than everyone else in the room. (There are a lot of things I'm not good at, but I'm good at my job, so say everyone I've worked with.)
I think one major difference is that I'm always studying, intentionally working to improve, every day. I've been doing that for twenty years.
I've worked with people who have "20 years of experience"; they've done the same job, in the same way, for 20 years. Their first month on the job they read the first half of "Databases for Dummies" and that's what they've been doing for 20 years. They never read the second half, and use Oracle database 18.0 exactly the same way they used Oracle Database 2.0 - and it was wrong 20 years ago too. So it's not just experience, it's 20 years of learning, getting better, every day. That's 7,305 days of improvement.
Programming is statistically a dead-end job. Why should anyone hone a dead-end skill that you won't be able to use for long? For whatever reason, the industry doesn't want old programmers.
Otherwise, I'd suggest longer training and education before they enter the industry. But that just narrows an already narrow window of use.
Table-ized A.I.
> The people can do both are smart enough to build their own company and compete with you.
Been there, done that. Learned a few lessons. Nowadays I work 9:30-4:30 for a very good, consistent paycheck and let some other "smart person" put in 75 hours a week dealing with hiring, managing people, corporate strategy, staying up on the competition, figuring out tax changes each year and getting taxes filed six times each year, the various state and local requirements, legal changes, contract hassles, etc, while hoping the company makes money this month so they can take a paycheck and lay their rent.
I learned that I'm good at creating software systems and I enjoy it. I don't enjoy all-nighters, partners being dickheads trying to pull out of a contract, or any of a thousand other things related to running a start-up business. I really enjoy a consistent, six-figure compensation package too.
Again, the quality applicant and the code monkey both have something the fakers do not-- Actual comprehension of what a program is, and how to create one.
As Bill points out, this is not the final exam. This is the "Oh, I see you do actually know how to program-- show me more" portion of the process. This is the part that HR drones are not capable of performing, due to Dunning-Krueger. Those that are actually, REALLY competent will do more than just satisfy the requirements of the challenge, they will provide actually working solutions to the challenge that properly validate their input, and return proper error states if the input is invalid, etc-- You can learn a LOT about a potential hire by observing their work. *THAT* is what this is really about. The triviality of the problem is a necessity, because you ***DON'T*** try to get free solutions out of people.
I realize that may be difficult for you to comprehend, but you *DON'T* do that. The job fair is to let people know that you have a position available, and try to curry interest in people to apply. A successful pre-screening is confidence building, and helps the potential hire to feel that your company is actually interested in actually hiring somebody, and not just fucking off in the booth, to cover for "failing to find somebody" and then "Getting yet another H1B". It gives them a chance to show you what they can do. That is what it is for, and what it does. It also excludes the fakers that this article is about-- The ones that can talk a good talk, but could not program a simple boolean check condition if their life depended on it.
If it were not for the time constraints of a job fair (usually only 2 days, and in that time you need to try and pre-screen as many as possible), I would suggest a tiered challenge, with progressively harder challenges, where you hand out resumes to the ones that make it to the top 3 brackets, but that is not the way the world works.
Kind of hard to take this article serious after saying gibberish like this. I would say most good programmers know that neither learn-to-code courses nor AI are going to make a dent in their income any time soon.
there is a huge shortage of decent programmers. I have personally witnessed more than one phone "interview" that went like "have you done this? what about this? do you know what this is? um, can you start Monday?" (120K-ish salary range)
partly because there are way more people who got their stupid ideas funded than good coders willing to stain their resume with that. partly because if you are funded, and cannot do all the required coding solo, here's your conundrum:
- top level hackers can afford to be really picky, so on one hand it's hard to get them interested, and if you could get that, they often want some ownership of the project. the plus side is that they are happy to work for lots of equity if they have faith in the idea, but that can be a huge "if".
- "good but not exceptional" senior engineers aren't usually going to be super happy, as they often have spouses and children and mortgages, so they'd favor job security over exciting ideas and startup lottery.
- that leaves you with fresh-out-of-college folks, which are really really a mixed bunch. some are actually already senior level of understanding without the experience, some are absolutely useless, with varying degrees in between, and there's no easy way to tell which is which early.
so the not-so-scrupulous folks realized what's going on, and launched multiple coding boot camps programmes, to essentially trick both the students into believing they can become a coder in a month or two, and also the prospective employers that said students are useful. so far it's been working, to a degree, in part because in such companies coding skill evaluation process is broken. but one can only hide their lack of value add for so long, even if they do manage to bluff their way into a job.
Comment removed based on user account deletion