What if all workers wrote software, not just the geek elite?
economist.comHaving taught at a coding bootcamp, I can tell you that some people will never in their lives be capable of writing functional software. People exist, with college degrees, who cannot in months of lessons, understand the idea of a variable.
Not every person in the world needs to or is capable of writing software. Probably a majority are given the right exposure but there's a big difference between being able to write any piece of code and creating business value with code.
I TA'd a lot of comp-sci classes in school - the double hump is real.
Even in a class designed to teach people basic, fundamental steps, some folks just will not engage with a computer in a meaningful way.
Not sure what it is, but intro CS classes will almost always show two bell curves at the end of the year - The double hump. The group that "got it" and the group that "did not".
You'll get downvoted for this, but it's been observed repeatedly.
It’s also the reason I stopped hiring from bootcamps. The signal to noise ratio is just too low.
What's your take on bootcamps? I worked pretty hard to get my BS CS, but now it seems like a lot of people I knew from college who were liberal arts majors are going to bootcamp. It's got me wondering about the value of my own degree.
Stuff they don't teach at bootcamps: Algorithmic complexity. "Don't roll out your own data structures" and "just use a library" works until it doesn't. Operating System fundamentals is another one.
It seems a lot of bootcamps teach with rote, rather than by looking at the underlying concepts. My favorite example is git. I've seen bootcamp grads claim they can use git, but what it really meant is that they memorized a few git commands and as long as they don't stray too fart from those they can sort of work using git. But cherry-picking, rebase, proper branching forget it.
I'm extremely skeptical of bootcamps, especially after learning that some of the TA's at Lambda (the most famous one) are hired to help with teaching as little as two months into the program as students[0]. I guess that counts toward their "placement" stats!
Not only that, but Lambda seems so desperate that they will offer a fresh grad at no cost to any company for a 4 week trial period. [1]
[0] https://nymag.com/intelligencer/2020/02/lambda-schools-job-p...
Interesting, thanks for the information. So it seems bootcamps are not really the same type of quality signal that you would get from a degree and GPA. This assuages my fears somewhat.
As an aside, I always considered git to be in the category of "Easily Google-able Technologies". In that whatever I am trying to do in git, it is generally very easy to find a resource online that tells me exactly what to do. For this reason I actually never really bothered to really "learn" git beyond git commit, git add, git push, etc. It's the same attitude I take towards technologies like CSS. Wanted to hear your thoughts on this as you seem experienced and I have < 3 YOE.
You should learn the basics of the git object model. It's really very simple and can help you understand better what some of the commands are actually doing. Lots of git stuff is just stored as plain text file inside the .git folder - the magic 'HEAD' ref is just a textfile that contains the hash of the HEAD commit.
http://shafiul.github.io/gitbook/1_the_git_object_model.html
Wow, this is really awesome. Thanks for the resource! Stuff like this is why I love HN.
> People exist, with college degrees, who cannot in months of lessons, understand the idea of a variable.
They were probably exposed to SICP and prefer functional programming. Lambda calculus, memory safe, concurrency safe :)
Its not like most people want to write software, but many people have to write software to keep the lights up in their home. Blame market of supply and demands that people are compelled to switch their jobs.
Many of my friends wanted to study physics, but due to no scope in employement, they are currently doing computer jobs and some excel related thing.
My experience is that most office workers are perfectly capable of writing software, especially if they have access to some basic training material and documentation.
However, most office workers would rather not write software and think that the whole pay-someone-else-to-do-it paradigm is great.
What if we ignored division of labor and asked everyone to be mildly capable at everything? Well, that sounds terrible.
It sure does!
On the other hand, having some mild programming capability gives a lot of people a lot more agency in their job and is a good thing. It's just that once they've proved out their idea and it is ready to be a real thing, they are overwhelmingly going to prefer to hand it off to the "experts".
I think there is a lot of opportunity in a no-code product that believes that this is how it works. I've just not ever seen anything where the developers say "this is great, I can slide right in and keep going with this!". It's always "what a mess, we have to start from scratch!"
I once knew of a person who needed to quickly come up with a report supporting a course of action some C-suite person wanted to do.
The report required access to some institutional financial data, which of course is "open" because this is a large non profit.
This clever person taught themselves enough code to ping the server for the data (read-only, of course). Note, this data is available to the business teams. The report was well done.
All hell broke loose organizationally, since they wanted strict access controls for security etc. The business team felt (perhaps justifiably) side-stepped (and perhaps unjustifiably threatened). Communications lines had been crossed! Access controls overcome! It was anarchy!
As a side note. Shame and ignorance-based access controls are a critical part of most orgs' fire-walling, I bet.
In my experience even IT people are not capable of doing that,sounds harsh but this is reality for me. IMHO.
I'd say many, not most, are capable of writing small scripts that save them time at work for small tasks.
It all falls apart when they try to turn it into software that does multiple things, has a GUI, is useful for anyone not using their specific machine setup a particular way, has error handling instead of checking the work output manually, etc.
I caught an interesting podcast this morning, about an artist who set out to create her own AI chatbot, with zero programming experience. She went to "a magical place called Github", where the software was provided along with a fair bit of assistance. The input was conversations with her family, which were presumably transcribed somehow. It's unclear how much programming she learned, but she did say that she spent 18 months "tweaking the algorithm".
She did figure out, one way or another, how to cobble together a microphone, speaker, and computer to make a talking chatbot. (Along with a 3D printed case to hold it.)
It wasn't especially good, but I found it remarkable that she was able to put all that together. I'm a professional programmer and that project would take me a while to pull off, just because I'd have to learn the specific technologies. I've actually wanted to do a Shakespeare-generating bot for a while, which should be an introductory project but is rather daunting to begin.
I really like the idea of more people programming. Most of what I do isn't difficult. In particular, I want to see less of a gap between "the programmers" and "the people who actually do stuff". As a programmer, perhaps my greatest skill is not the languages and frameworks and debugging tactics, but my ability to talk to the people who need stuff and quickly learn enough of their needs to help. It would be even better if they could do a lot of it themselves, and bring me in for the hard parts.
I'm semi-technical & worked mostly in non dev roles. I've been coding apps all the time in my previous jobs. Mostly this was done in JIRA or Excel - creating a workflow, validating new data that comes in. Easy repetitive things that frustrated everyone on the team & myself the most. Now I am helping other to do this as the co-founder of a no-code SaaS (Chartmat). the amount of people who love to build without depending on devs is mind-blowing. I'm looking forward to this next decade in which building websites, dashboards and apps becomes more of a commodity that everyone can do and devs move on to more complex and critical functions such as writing smart contract & scaling apps.
Assuming you want to use Python or JavaScript why not ?
It's much safer than working on your own car.
Hell it's even safer than fixing a bike.
I would never impose this on people though. Many people just don't want to do it.
It depends what outputs you drive with your JS or Python code.
Then the "geek elite" would spend more time fixing things they should have learned before starting. Devs who have fixed general office worker's abominations of Microsoft Access and turned them into proper apps would agree. This isn't to belittle the creations, just an observation. The work the normal office workers build had value to the business or their managers wouldn't have let them build it. The issue comes in risk to the business when best practices are unknown and not followed. When the Access app collapses under its own weight is often when "geek elite" are summoned to pull the team's mission critical homegrown application out of the fire, and the expectations are often not matched when they build something the wrong way it takes a few days and it might take a few sprints to "do it right." The solution is probably to meet in the middle and take the apps made by the regular office workers and migrate the ones with value as soon as possible. Accruing technical debt is cheaper and it can be unpaid until disaster strikes so that is often the choice of management especially if the environment or relationship between front line office workers and internal app developers is strained.
The geek elite? What is this, 1990?
The 1990s had the "Digerati" (term coined in '92).
So long as they support and maintain their own code and accept responsibility for it.
If the software does the right thing for the person or organisation then it's time for the pros to come in and rewrite it all pretty and snappy.
Discovering of what software should do is like half of the development. I don't mind it being written by the people who are closer to the matter at hand.
I disagree, maybe they want it to do more but they don't know it is possible. Heck, I can code just fine but some problems I need to ask a proper software dev if they know the right algorithm/ datastructure,etc... to make it possible. And don't even get me started with security issues that arise when people are hard coding creds and screwing up permissions on prod data.
Quite a few "non-programmer" workers do indeed write software already, and have been doing so for decades now. The only trouble is that said software exists in the form of Excel spreadsheets and Access databases ;)
It'd be interesting to see such "programming-but-it-doesn't-really-feel-like-programming" environments get generalized a bit into something that can produce proper general-purpose applications (preferably in a form that's actually friendly to sensible version control systems).
The interface may look cluttered: the landing
page jams in 150 buttons and a local-news ticker
I think the next big thing in software is going to be that almost every office worker will start writing basic software. Maybe in a python like language.
People spend a lot of time and effort to do simple repetitive stuff that anybody can do with a 10 line python script. They don't do it because nobody has told them it's that easy.
30-40 years ago, an average office worker wouldn't use a word processor for typing out documents. They would have probably used a typewriter or written it by hand. Now everybody uses Microsoft Word. And they use it well, without major issues.
The same thing will happen to python or another language like it.
At my current work I use linux and bash to rename files, filter csv data, and basic summary statistics why? one day I asked a colleague of mine to count the number of occurrences of a persons name in a google sheet...watched him count using his mouse.
I'd like for people to get to python levels of coding no matter how light, but baby steps first.
Archived: https://archive.ph/FtdBG