9 ms·
The topic of your rant is something that it took me forever to accept as a software engineer. Nobody cares about quality. Your method is refactored to be super
by ryandrake 9y ago
The topic of your rant is something that it took me forever to accept as a software engineer. Nobody cares about quality.
Your method is refactored to be super-fast and have no branches? Nobody cares.
Elegant, re-usable architecture? Nobody cares.
You've optimized an inner loop in assembly and increased speed by 30%? Nobody cares.
Made it to zero compiler warnings? Nobody cares.
No known crashers left in the bug tracker? Nobody cares.
Fixed that bug that's been in the code for two years? Nobody cares.
Patched a security hole that could allow a crazy bad attack? Nobody cares.
The vast majority of software shops out there care only that something--anything--ships as fast as remotely possible, and as long as it does not kill the customer, it's a success. Nobody in leadership gives a single fuck about anything else. There are a few places where this is not true, and good software people tend to collect there, but it's true for the vast majority of places where you will work. The sooner you accept this, the less likely you'll burn out.
- hcurtiss 9y agoIf there were a list of such companies, I would patronize them.
- whatshisface 9y agoMy universal advice to graduating CS students is, "no matter what your current code quality/dev speed balance is, your boss will think you're polishing too much."
- user5994461 9y agoSecond advice. If you want to pretend that quality matters, move to aerospace, finance, or any critical systems. Third advice. It's not enough to fix bug, everyone MUST know that you fixed that weird bug that customers were complaining about for years and noone could understand the cause.
- adrianN 9y ago> If you want to pretend that quality matters, move to aerospace, finance, or any critical systems. The important word here is pretend. If the system is safety critical you'd think that quality matters. It doesn't. What matters is following the process laid out in ISO 26262 or whatever standard applies to your field. Whether or not that improves quality is not important (it does, but only a little). The important part is that you have reduced/no liability for accidents if you can prove that you did everything by the book.
- andai 9y agoandai.depression++;
- daemin 9y agoYep, part of the quality standard for medical devices is to have processes, and also processes to improve and change the processes. So theoretically there shouldn't be any doubt as what to do in a given circumstance when something fails, and that standards would rise as development matures. But in reality it's just a bunch of paperwork, and as far as the software is concerned, standard practices of unit and integration testing, code reviews, and strict compiler settings are more than adequate already. Edit: It's all about documenting that those things have been done and signed off on, so keeping track of paperwork.
- user5994461 9y agoTesting, review, history of changes... many things that are completely unheard of in the average small development shop.
- tzahola 9y agoExactly. A prime example of this is the “quality” software at financial companies.
- user5994461 9y agoThey care about loosing money and risking to lose money.
- metaobject 9y agoAnd make sure to record it so that at the end of your yearly review cycle, and before the raises are decided, you can remind your upper management of your notable accomplishments.
- 77pt77 9y agoEither it's easily perceived as bringing value or it won't have any effect. If you have to remind people of it, it's as good as not existing in the first place.
- taejo 9y agoI disagree. A year is a long time, and a valuable, difficult achievement in February may raise your prestige for a while, but by December that's faded to background levels. I even forget my own achievements from nine months ago, never mind those of a dozen other people on my team.
- 77pt77 9y agoThe question is always: > What can you do for me or have done for me recently? Services provided so far in the past that need to be reminded are at best irrelevant and it can actually hurt you to bring them up. I'm not saying that it is nice, only that humans behave individually and (even more collectively) according to this.
- taejo 9y agoSo at my annual review in December I should only be assessed on work in the last quarter? I guess I'd better save my best work for November.
- webmaven 9y agoDepending on the attention span of your particular managers (and their managers as well, most likely), this can indeed be a viable strategy. You might also carefully consider your company's culture when deciding how to split your efforts between new (user-visible) features, preventive maintenance, and bugs affecting end users (especially any VIP end users that have your managers' ears). For example, a 'Hero' culture rewards last-minute, late night, death march efforts, whether to fix an urgent problem, or to add that one feature that is (supposedly) needed to land a specific high-value customer/sale.
- flukus 9y ago> The vast majority of software shops out there care only that something--anything--ships as fast as remotely possible When nobody cares about you get the point sooner or later where you can't ship, they've probably launched a product but there's still a lot more to ship in future. Eventually they become so mired in technical debt that they care. When they realize they've invested a hundred million dollars in something that now has no value? They care. (especially if it's a turd before it gets out the door) When a start up comes along and eats their lunch because they aren't weighed down be legacy code? They care. When they lose revenue because they can't keep customers happy? They care. The problem is that the professionalism of software developers doesn't allow them to learn these lessons early and hard enough.
- killjoywashere 9y ago> They care. > They care. > They care. > They care. I disagree. They will hire a consulting company to do a root cause analysis. The Ha'vad grads from McKinsey will swoop in, determine that the past doesn't matter, you need to get to black and to get to black, the St. Louis team needs to go, along with 20% of the following 7 teams.... No, they don't care. And never will.
- nickthemagicman 9y agoThen they just rewrite it again shittily. lol
- QuackingJimbo 9y ago> There are a few places where this is not true Where?
- aswanson 9y agoProbably the Apollo moon lander software team, circa 1968.
- avs733 9y agohttps://www.fastcompany.com/28121/they-write-right-stuff https://www.fastcompany.com/28121/they-write-right-stuff
- wyager 9y agoThis article bothers me because it implies that the path to correct code is simply to have code written by white-collar salarymen (as opposed to unkempt youth), and not improved formal methods.
- avs733 9y agoIt doesn't imply there is a correct path to code. It implies that the best path to the goal for this program is to minimize the opportunity for exhaustion to be a variable.
- andai 9y ago> That’s the culture: the on-board shuttle group produces grown-up software, and the way they do it is by being grown-ups. It may not be sexy, it may not be a coding ego-trip — but it is the future of software. When you’re ready to take the next step — when you have to write perfect software instead of software that’s just good enough — then it’s time to grow up.
- gaius 9y agobut it is the future of software It's over two decades since that was written. I remember reading it when it was first published. Does it look much like the future?
- eric_b 9y agoI agree with you, and this is a hard lesson to learn. But I don't think it's necessarily a bad thing. Software should really only be measured by the value it provides. If a terribly buggy piece of poorly written code still saves hundreds of man hours a week, it's a win. (Unit tests be damned) If that big ball of mud that's using 20 year old technology still prints a billion dollars a year - that's a win. (Microservices be damned) If some slapdash jquery-and-duct-tape web app still solves a specific problem I have, that's a win. (React SPAs be damned) Does crappy software flourish while beautiful software dies? Sometimes. But usually code hygiene, security, bug counts and crashes are not what makes or breaks the success of a piece of software. Developers often think their work is the most important, when in actuality it is usually not. Sometimes you really just are a cog in the machine.
- mtpn 9y agoI’m a web developer in my spare time. I write little things that solve problems in my two day jobs, they haven’t tended to be pretty because often I’m learning as I go and when it works then it’s truly time to move on to the next problem, even though I now know enough to solve the old problem again more elegantly. I’m trying to use better practices and modern JS frameworks in new projects just for experience, even though they are actually overkill in many cases. Certainly the people I work for would rather have two useful utilities written in plain es5 and php than one useful utility plus a story about the code quality. I’m gradually finding the middle ground where somebody call look at my code and not instantly want to take a shower.
- andrei_says_ 9y agoSaying all this shows that 1. You care and 2. You know the difference. I’d hire you.
- mtpn 9y agoGood to hear- I’m hoping to switch to full time web development some time in the new year. It's actually my favorite thing at these other jobs.
- tlear 9y agoReally? Ship an app to the store that crashes all the time and see your metrics go WAY down. Loosing user data is another great way to lose users permanently. Sure if you are Google of FB it probably does not matter but for smaller companies NOT CRASH, NOT LOOSE USER DATA seems to matter quite a bit in my experience.
- jdietrich 9y agoB2C is a tiny, tiny fraction of the overall software industry. The vast majority of developers are working on godawful enterprise software that end-users have no choice but to tolerate. The absolute worst apps on the iOS store are remarkably high quality by the standards of enterprise software. I suggest reading the comments on this thread: https://news.ycombinator.com/item?id=13859961 https://news.ycombinator.com/item?id=13859961
- TuringNYC 9y agoSpot on in my experience. Curious if anyone at Google/Facebook/Amazon/Apple core teams could comment on whether these issues are also the case at Google/Facebook/Amazon/Apple? I would imagine solving issues like these do make business sense at big firms with scalability problems.
- mch82 9y agoThis will be changing as software is responsible for the control of an increasing number of physical systems (cars, medical, aircraft, etc.) Edit: Unless we elect people who take away legal protections in place to regulate safety critical software
- ehnto 9y agoI had the fortune to work at an agency that cared a whole lot, and tended to take on clients who had learned hard lessons from bad code written for the platform we worked in. Code quality matters and has tangible consequences in that niche, and it was awesome to work in that environment. I haven't worked anywhere like that previously, and I think the key factors that brought code quality to the forefront were: * Bugs cost sales * Poor code caused friction for features a significant amount (reduced ROI) * Small teams and an insanely tight customer relationship to those teams. The client needed to have a stakeholder in our team or as close to that as possible. But most importantly it was just the company culture from top to bottom. The director cared, he made the project managers care, and we already cared because it's more fun to write good code.
- egisspegis 9y agoThat's a nice place to work in! How was this company doing? Would it be doing better just shipping "if it makes money - it's good enough" code?
- ehnto 9y agoIt really was! I believe they were doing well, growing at a steady pace. Another part of the puzzle was that client relationships were long, over years, so if we wrote bad code it was us that would have to deal with it later. The clients were typically larger, which helped to that end. We did take on a rushed project at one point, so there is a good comparison in that project. It shipped, but everyone was unhappy, some key people left and had to be re-hired for. It triggered a very transparent discussion roughly titled "let's figure out how to never do that again" and to that end they curated the clients more carefully, and put in some checks to see if projects are a good fit. But I understand that is probably unique, and at previous jobs the "good enough to make money" mantra has worked out okay. I think when your software domain is complex and rapid, it pays to have sound architecture. When it is simple, or slow to change, it doesn't matter quite as much.
- gaius 9y agoThe industry does itself no favours by regularly rejecting techniques to improve software quality tho'. We could all be writing rock solid Ada, but programmers said no, we prefer JavaScript, hell we'll even try to use it on the server...
- barrkel 9y agoLack of profit focus is definitely something I see in junior devs of the higher quality kind. It's important to sell any prospective improvement to management in a way that makes money. Quality people can't see won't make money, and even quality they can see will only make money if there's a competitive market.
- rimliu 9y agoDo they care if a year down the road your maintenance costs increase fivefold? Have they seen the analysis which shows that initial development cost just makes a small fraction of the total cost and that fraction get smaller the longer software is in use? Do they care if a year down the road your TTM times increase tenfold? We know that the bad code slows us down. Why do we write it? Why do we write the code that slows us down? How much does it slow us down? It slows us down a lot. And every time we touch the modules we get slowed down again. Every time we touch bad code it slows us down, what do we write this stuff what slows us down? And the answer to that is: well we have to go fast. I'll let you to deal with the logical inconsistency there, my message to you today is—you don't go fast by writing crap. https://www.youtube.com/watch?v=TMuno5RZNeE&feature=youtu.be&t=956 https://www.youtube.com/watch?v=TMuno5RZNeE&feature=youtu.be...
- wellboy 9y agoNo. When they have seen all the bugs, they will tell you to fix them and when you make a new feature they will tell you the same thing, just make it work, just ship it.
- walshemj 9y agoif you improved the efficiency of a hpc shot by 30% I am sure people would sit up and take notice :-)
- otakucode 9y ago>as long as it does not kill the customer Whoa now. The Toyota 'unintended acceleration' case showed us that it's fine to kill the customer. The court declared themselves incapable of penalizing even the most stupendously egregious example of criminal negligence we're likely to see (Of 90+ 'required' and 'suggested' practices in automotive industry firmware coding practices, they did 4 in the code in question. The developers didn't have a bug tracker or static analysis tools. Typical of all organizations, the software engineers had no real power to demand more time or testing, etc). They lost a civil wrongful death case and paid out $4 million, then settled with the plaintiffs out of court before the jury could award punitive damages. So killing is OK. The issue is that, I think, software people need to stand up for themselves. They need to tell executives no, and refuse to pump out slapdash work. They need to take responsibility for the successes and failures of the company. The practical truth is that in most organizations, the software is the most valuable part of the entire business and the thing with the power to enable the company to excel or fall flat. We've been waiting for executives to recognize this since the 1980s and they still just see the software people as people who do typing.
- wfo 9y agoTotally agree. But the only way for software people to take any power or control (really, for any worker to have any say in the business at all) is to unionize, and us tech folks tend to treat that like a curse word.
- ryanmarsh 9y agoUnion is a curse word but I have a hunch that there are others that might make more sense. Guild maybe?
- bhk 9y agoBad example. After a 10-month search, NASA and NHTSA scientists found no electronic defect in Toyota vehicles. Driver error or pedal misapplication was found responsible for most of the incidents. The report ended stating, "Our conclusion is Toyota's problems were mechanical, not electrical." https://en.wikipedia.org/wiki/2009%E2%80%9311_Toyota_vehicle_recalls https://en.wikipedia.org/wiki/2009%E2%80%9311_Toyota_vehicle...
- cr0sh 9y ago> The sooner you accept this, the less likely you'll burn out. I learned this at my first software engineer (though we didn't call it such) job back when I was 18; closing in on 45 years of age, and I'm still here at it, and don't see that changing soon. I have thought about management, but I really don't have such experience, and I'm not a real people person anyhow - plus I like to code. Another thing I have learned - at least here in the United States: Not only do they not care about quality, they usually don't have any loyalty to you as an employee. They'll let you go without any warning at all, and if you're lucky, you'll get some severance pay. So save your money, build a savings account with FU cash in it (6 months to a year of salary - or more), and don't be afraid to drop an employer like a hot potato if things aren't working out, or a better offer is in the works. And above all else, don't let your skills stagnate.