5 ms·
> Worried that your job is in danger of being automated away by software? Learning to code could be a useful insurance policy. I often worry that we're setting
by sanderjd 11y ago
> Worried that your job is in danger of being automated away by software? Learning to code could be a useful insurance policy.
I often worry that we're setting ourselves up for major failure here. There is no fundamental principle that suggests most kinds of programming are any less susceptible to automation than any other kind of work. Especially the kinds of things that are taught in most "learn to code" programs – simple web applications – are increasingly commoditized or able to be built out of commodity pieces, and that trend will only increase. Another popular reason to learn to code is to do data analysis or procedural tasks for your line of business, but in the long run, those tasks are probably better served by more purpose-built tools that have more in common with spreadsheets than the types of things professional software developers do.
I guess my point is that there seems to be a meme that extolls the virtues of "learning to code" as an end, rather than as a means to be more effective at something else. I'd rather see articles about what those somethings-else are, and how knowing how to program can augment them, but those seem fewer and further between.
- outworlder 11y ago> There is no fundamental principle that suggests most kinds of programming are any less susceptible to automation than any other kind of work Well, as a fundamental principle, there's the Halting Problem. The simplest programming tasks will be simplified to some extend or be commoditised. In much of the same way we can purchase website templates and website builders can get the trivial sites up and running with no knowledge. In the past, HTML was required knowledge to publish a blog, but not anymore. But there's a reason most professional applications are not developed with such tools. Given that, I expect the careers of programmers that aren't IDE-pilots are safe for the foreseeable future. Increasingly complex task will be, and already are, being automated. But so far, all that progress has only increased the programmer's leverage. > I guess my point is that there seems to be a meme that extolls the virtues of "learning to code" as an end, rather than as a means to be more effective at something else. I completely agree with that. Outside of IT, there are many areas that would benefit from some programming knowledge and are currently under-served.
- wuschel 11y ago> I completely agree with that. Outside of IT, there are many areas that would benefit from some programming knowledge and are currently under-served. I'd like to emphasize this sentence. There is not just binary code that is used in software. Great challenges arise in other domains e.g. such as synthetic biology.
- sanderjd 11y agoThat's exactly my point: I want to hear less "learn to code", and more "learn xyz for which programming is useful". My sense is that being a professional programmer is much more of a no-brainer right now than those xyzs, but that it is very short-sighted as a society-wide meme. I want to see software create a boom in jobs outside the creation of software.
- sanderjd 11y agoYour appeal to the Halting Problem is basically why I included the "most kinds of" weasel words. As much as we don't like to admit it, the vast majority of professional programmers are doing things that are properly categorized as "the simplest programming tasks". You're right that so far the automation trend has only increased programmers' leverage, but I don't think there's any more reason to think that will be an indefinite trend than there was to think automotive assembly would always be a high-skill job. I think programming should be thought of as a useful skill (like math or writing) that for most people comes in handy in small but myriad ways, while a smaller segment of people gain more expertise and do it professionally (like mathematicians or writers). Maybe that's how most of this nascent movement is thinking of it, but it seems to me it's being sold more as a solution to the future job woes that many foresee.
- Lawtonfogle 11y ago>I don't think there's any more reason to think that will be an indefinite trend than there was to think automotive assembly would always be a high-skill job. It isn't. One day even programmers will, in general, be replaced. The way I see it is that programmers will be one of the last ones to be replaced because once they are replaced, everything else that can be replaced that isn't will soon follow (as replacing programmers will send the process of replacing such jobs into overdrive). Everyone learning programming does not solve the bigger problem for society, but it is probably one of the safest careers even if it isn't safe.
- wslh 11y ago> I often worry that we're setting ourselves up for major failure here. There is no fundamental principle that suggests most kinds of programming are any less susceptible to automation than any other kind of work. I think there is some kind of denial in our community about this. A decade ago I (like anyone else) hired web designers for every iteration of my company web site. Now I use Bootstrap and give the designer some small tasks to customize it. More than a decade ago there was SOAP to make web services and then JSON came up and simplify the web service ecosystem. We can say that there will be no work for simple development tasks, only work for more complex things, that eventually will be part of some future product.
- sanderjd 11y agoI don't think it's denial, because it has simply been true that even as many tasks have been automated away, there have been multiplicatively more tasks that haven't yet. Nobody really knows whether we should be optimistic or pessimistic about that trend continuing, but I think we should be more aware that it isn't fundamentally different than similar trends in other types of work in the past.
- wslh 11y ago> it has simply been true that even as many tasks have been automated away, there have been multiplicatively more tasks that haven't yet. In part this is the denial I talked about. Many of the multiplicatively tasks we have are the result of not finding how to solve obvious problems with pragmatic solutions. Let me explain: Everytime we came up with new programming environments and languages we repeat the same cycles. It is very rare to have a real breakthrough. For example, once NodeJS become popular we are happy when someone introduce libraries that already existed in another language instead of automating the production of these libraries. All this work can be reduced to some software to generate stuff for multiple languages. Surely you can't apply this for different paradigms but there are few programming paradigms being used in production. If you follow this approach you will only need a proportion of the people that is working now. Another example: big corporations. I don't talk about Google or Facebook, I talk about the typical big corporation. If you walk around you can find that replacing a lot of IT employees is relatively easy with good organization. In a way we are protecting our own work not thinking very hard about how to replace people.
- unabst 11y agoWhether coding is a means or an end can be interchanged once one can code, so the importance of coding as a general skill still stands. And just from personal experience and observation, I have seen friends generate work for themselves (granted, at where they work) simply by being able to code. They would be asked to fix the web site or modify a spreadsheet macro, and they'd be able to do it. What coding has going for it: 1. Coding is higher up the abstraction stack. The higher the safer. Lower would be performing concrete instructions like picking berries or mopping floors. The more abstract the work, the harder it is to be automated. Because... 2. Computers can compile but still cannot abstract. Most abstractions embody arbitrary structures external to the computer. So no matter how smart a computer gets, people will still need to communicate these wants. There will always be a place for the translator, and we might as well call this the coder. 3. There is still so much room for innovation in the way we choose our abstractions and build our structures. There are human trends, technological trends, and design trends, making software a constantly moving, upgrading, ephemeral target. There is still plenty of potential "disruption" at every scale, and coders can be a part of it. There will always be a place for the architect, aka coder. 4. With more services exposing APIs and more programs being scriptable, it is becoming easier for coders to apply themselves in any logistical situation to hook things up and to further automate. This is the plumber, aka coder. Granted, the above is only why coding is comparatively a more useful skill than other skills. Coding can still be extremely hard to acquire for some people. There will always be better coders than you. And, if everyone can code, the skill itself will lose value. So though coding may not be automated like everything else, whether it will save anyone is likely to be a different story entirely.
- sanderjd 11y agoGreat comment, thanks. I mentioned in a couple other comments here that I'm very much a believer in programming as a skill that augments other work. I would just like to see a boom in that sort of work, rather than lots of people training up for pure software jobs. If a smart 18-year-old is trying to figure out what to do with their lives, I would rather society tell them to, eg. become a biologist and make sure to take some good programming classes, rather than to learn how to make mobile apps.
- spott 11y ago>There is no fundamental principle that suggests most kinds of programming are any less susceptible to automation than any other kind of work. There are two ways to "automate" programming. Make the language/framework easier to understand/write in, which isn't fundamentally different than it is now, or make the computer understand natural language and do interpretations on its own. The first is likely to happen, but learning to code now will only make you better at writing in whatever simplified language is developed next. The second is AI complete and won't happen till we develop REAL AI... which we aren't really that close to doing. Learning to code is just learning to be very precise with your instructions. That is a useful skill regardless of whether you are actually coding or not.
- sanderjd 11y agoWhat I'm thinking of are a subset of your first thing: powerful task-specific interfaces that act somewhat like DSLs, but usually come with a sophisticated GUI. Some examples: Excel, Photoshop, AutoCAD, Ableton. I think (and hope) these sorts of solutions will proliferate and a bunch of good ones could make a major dent in the world's need for people who program professionally. I think the "we won't be automated away until REAL AI happens" thing is a trap. Certainly it's true that there will be some programmers working professionally until that point, but I see no reason to expect that it will be a huge number. I wish I could agree with you more that "learning to code is just learning to be very precise with your instructions". I'm sure there are some programs that stress that, but most of the things I see (especially those targeted at adults) are more focused on the minutia of how we actually write professional software in 2015, because they are targeting job re-training rather than the acquisition of a skill that is broadly useful. In summary, I'm just not convinced that "learning to code could be a useful insurance policy" is the right message to be sending. Maybe I'd be on board with something more like, "learning to code is probably better than the alternatives", but that leads me to think that we could use some more alternatives.