Friday, December 28, 2012

State of the New Machines

I'm typing in this post from a cottage in northern Ontario. This is my first post on my iPad, connect via 3G to the world. I couldn't imagine 20 years ago, while lugging home my 30 pound Compaq "portable", that someday I'd be miles off the grid, yet connect to the world via a wafer thin screen resting on my laptop. Hardware, it seems, has made amazing progress.

The irony is probably that my tablet is packed with retro apps, most of which are throw backs to the early days of PC computing. While hardware has screamed ahead, software has, well, lagged. Still, there are a few apps that impress me, mostly the stuff coming from Autodesk. The rest however...

Sunday, November 25, 2012

Theory and Practice

Nearly three decades ago, when I started university all I really wanted to learn was the magic of programming. But my course load included plenty of mathematics and computer theory courses, as well as crazy electives. “What does all this have to do with programming?” I often complained. At first I just wished they’d drop the courses from the curriculum and give me more intensive programming assignments. That’s what I thought I needed to know. In time I realized that most of it was quite useful.

Theory is the backbone of software development work. For a lot of programming tasks you can ignore the theory and just scratch out your own eclectic way of handling the problem, but a strong theoretical background not only makes the work easier it also is more likely to withstand the rigors of the real world. Too often I’ve seen programmers roll their own dysfunctional code to a theoretical problem without first getting a true appreciation of the underlying knowledge. What most often happens is that they flail away at the code, unable to get it to be stable enough to work. If they understood the theory however, not only is the code shorter, but they’d spend way less time banging at it. It makes it easier. Thus for some types of programming, understanding the underlying theory is mandatory. Yes, it’s a small minority of the time, but it’s often the core of the system, where even littlest of problems can be hugely time intensive.

The best known theoretical problem is the ‘halting problem’. Loosely stated, it is impossible to write some code that can determine if some other code will converge on an answer or run forever (however one can write an estimation that works with a finite subset within a Turing Machine and that seems doable).  

In its native form the halting problem isn’t crossed often in practice, but we do see it in other ways. First is that an unbounded loop could run forever. An unbounded recursion can run forever as well. Thus in practice we really don’t want code that is ever unbounded -- infinite loops annoy users and waste resources -- at some point the code has to process a finite set of discrete objects and then terminate. If that isn’t possible, then some protective form of constraint is necessary (although the size should be easily configurable at operational time).

The second way we see it is that we can’t always write code to understand what code is trying to do. In an offbeat way, that limits the types of tools we can use in automation. It would be nice for instance if we could write something that would list out the stack for all possible exceptions in the code with respect to input, but that would require the lister to ‘intelligently’ understand the code enough to know the behavior. We could approx that, but the lack of accuracy might negate the value of the tool.

Another interesting theoretical problem is the Two Generals Problem. This is really just a coordination issue between any two independent entities (computers, threads, processes, etc.). There is no known way to reliability get 100% communication if the entities are independent. You can reduce the window of problems down to a tiny number of instructions, but you can never remove it entirely. With modern computers we can do billions of things within fractions of a second, so even a tiny 2 ms window could result in bugs occurring monthly in a system with a massive number of transactions. Thus what seems like an unlikely occurrence can often turn into a recurring nuisance that irritates everyone.

Locking is closely related to the Two Generals Problem. I’ve seen more bugs in locking than in any other area of modern programming (dangling pointers in C were extremely common in the mid 90s but modern languages mitigated that). It’s not hard to write code to lock resources, but it is very easy to get it wrong. At its heart, it really falls back to a simple principle: to get reliable locking you need a ‘test-and-set’ primitive. That is, in one single uninterrupted single-threaded protected operation, you need to test a variable and set it to ‘taken’ or return that is it unavailable. Once you have that primitive, you can build all other locking mechanisms on top of it. If it’s not atomic however, there will always be a window of failure. That links back to the Two Generals Problem quite nicely, since where it becomes an issue is when you can’t have access to an atomic ‘test-and-set’ primitive (and thus there will always be problems).

Parsing is one of those areas where people often tread carelessly without a theoretical background, and it always ends badly. If you understand the theory and have read works like The Red Dragon Book then belting out a parser is basically a time problem. You just decide what the ‘language’ requires such as LR(1), and how big the language is and then you do the appropriate work, which more often than not is either a recursive descent parser or a table driven one (using tools like lex/yacc or antlr). There are messy bits of course, particularly if you are trying to draft your own new language, but the space is well explored and well documented. In practice however what you see is a lot of crude split/join based top-down disasters, with the occasional regular expression disaster thrown in for fun. Both of those techniques can work with really simple grammars, but then fail miserably when applied to more complex ones. Thus being able to parse a CSV file, does mean you know how to parse something more complex. Bad parsing usually is a huge time sink, and if it’s way off then the only reasonable option is to rewrite it properly. Sometimes it’s just not fixable.

One of my favorite theoretical problems is the rather well-known P vs NP problem. While the verdict is still outstanding on the relationship, it has a huge implication for code optimizations. For people unfamiliar with ‘complexity’, it is really a question of growth. If you have an algorithm that takes 3 seconds to run with 200 inputs, what happens when you give it 400 inputs? With a simple linear algorithm it takes 6 seconds to run. Some algorithms perform worse, so they may take 9 secs (3^2 -- three squared) to run, or even 64 seconds (4^3 -- four to the power of three). We can take any algorithm and calculate its ‘computational complexity’ which will tell us exactly how the time grows with respect to the size of the input. We usually categorize this by the dominant operators so O(1) is a constant growth, O(n) is growing linearly by the size of the input, O(n^c) is growing by a constant exponent (polynomial time) and O(c^n) has the size of the input as the exponent (exponential time). The P in the equation is a reference to polynomial time, while NP is rather loosely any growth such as exponential that is larger (I know, that is a gross oversimplification of NP, but it serves well enough to explain that it references problems that are larger, without getting into what constrains NP itself).

Growth is a really important factor when it comes to designing systems that run efficiently. Ultimately what we’d like is to build is a well-behaved system that runs in testing on a subset of the data, and then to know when it goes into production that the performance characteristics have not changed. The system shouldn’t suddenly grind to a halt when it is being accessed by a real number of users, with a real amount of data. What we’ve learned over the years is that it is really easy to write code where this will happen, so often to get the big industrial stuff working, we have to spend a significant amount of time optimizing the code to perform properly. The work a system has to do is fixed, so the best we can do is find approaches to preserve and reuse the work (memoization) as much as possible. Optimizing code, after its been shown to work, is often crucial to achieving the requirements.

What P != NP is really saying in practice is that there is a very strong bound on just exactly how optimized the code can really be. If it’s not true then there would be no possible way you could take an exponential problem and find clever tricks to get it to run in polynomial time. You can always optimize code, but there might be a physical bound on exactly how fast you can get it. A lot of this work was best explored with respect to sorting and searching, but for large systems it is essential to really understand it if you are going to get good results.

if it were true however, amongst many other implications, that would mean that we are able to calculate some pretty incredible stuff. Moore’s law has always been giving us more hardware to play with, but users have kept pace and are continually asking for processing beyond our current limits. Without that fixed boundary as a limitation, we could write systems that make our modern behemoth's look crude and flaky, and it would require a tiny fraction of the huge effort we put in right now to build them (also it would take a lot of fun out of mathematics according to Gödel).

Memoization as a technique is best known from ‘caching’. Somewhere along the way, caching became the over-popular silver bullet for all performance problems. Caching in essence is simple, but there is significant more depth there than most people realize, and as such it is not uncommon to see systems that are deploying erratic caching to harmful effect. Instead of magically fixing the performance problems, they manage to make them worse and provide a slew of inconsistencies in the results. So you get really stale data, or a collection of data with parts out of sync, slower performance, rampant memory leaks, or just sudden scary freezes in the code that seem unexplainable. Caching, like memory management, threads and pointers is one of those places where ignoring the underlying known concepts is most likely to result in pain, rather than a successful piece of code.

I’m sure there are plenty of other examples. Often when I split programming between ‘systems programming’ and ‘applications programming’ what I am really referring too is that the systems variety requires a decent understanding of the underlying theories. Applications programming needs an understanding of the domain problems, but they can often be documented and passed on to the programmer. For the systems work, the programmer has to really understand what they are writing, for if they don’t, the chances of just randomly striking it lucky and getting the code to work are are nearly infinitesimal. Thus, as I found out over the years, all of those early theory courses that they made me take are actually crucial to being able to build big industrial strength systems. You can always build on someone else’s knowledge, which is fine, but if you dare tread into any deep work, then you need to take it very seriously and do the appropriate homework. I’ve seen a lot of programmers fail to grok that and suffer horribly for their hubris.

Sunday, November 18, 2012

Best Practices

One significant problem in software development is not being able to end an argument by pointing to an official reference. Veteran developers acquire considerable knowledge about ‘best practices’ in their careers, but there is no authoritative source for all of this learning. There is no way to know whether a style, technique, approach, algorithm, etc. is well-known, or just a quirk of a very small number of programmers.

I have heard a wide range of different things referred to as best practices, so it’s not unusual to have someone claim that their eclectic practice is more widely adapted than it is. In a sense there is no ‘normal’ in programming, there is such a wide diversification of knowledge and approaches, but there are clearly ways of working that consistently produce better results. Over time we should be converging on a stronger understanding, rather than just continually retrying every possible permutation.

Our not having a standard base of knowledge makes it easier for people from outside the industry to make “claims” of understanding how to develop software. If for instance you can’t point to a reference that says there should be separate development, test and production environments, then it is really hard to talk people out of just using one environment and hacking at it directly. A newbie manager can easily dismiss 3 environments as being too costly and there is no way to convince them otherwise. No doubt it is possible get do everything all on the same machine, it’s just that the chaos is going to extract a serious toll in time and quality, but to people unfamiliar with software development issues like ‘quality’ find that they are not easily digestible.

Another example is that I’ve seen essentials like source code control set up in all manner of weird arrangements, yet most of these variations provide ‘less’ support than the technology can really offer. A well-organized repository not only helps synchronise multiple people, but it also provides insurance for existing releases. Replicating a bug in development is a huge step in being able to fix it, and basing that work on the certainty that the source code is identical between the different environments is crucial.

Schemas in relational databases are another classic area where people easily and often deviate from reasonable usage, and either claim their missteps as known or dismiss the idea that there is only a small window of reasonable ways to set up databases. If you use an RDBMS correctly it is a strong, stable technology. If you don’t, then it becomes a black hole of problems. A normalized schema is easily sharable between different systems, while a quirky one is implicitly tied to a very specific code base. It makes little sense to utilize a sharable resource in a way that isn’t sharable.

Documentation and design are two other areas where people often have very eclectic practices. Given the increasing time-pressures of the industry, there is a wide range of approaches happening out there that swing from ‘none’ to ‘way over the top’, with a lot of developers believing that one extreme or the other is best. Neither too much or too little documentation serves the development, and often documentation isn’t really the end-product, but just necessary steps in a long chain of work that eventually culminates in a version of the system. A complete lack of design is a reliable way to create a ball of mud, but overdoing it can burn resources and lead to serious over-engineering.

Extreme positions are common elsewhere in software as well. I’ve always figured that in their zeal to over-simplify, many people have settled on their own unique minimal subset of black and white rules, but often the underlying problems are really trade-offs that require subtle balancing instead. I’ll often see people crediting K.I.S.S (keep it simple stupid) as the basis for some over-the-top complexity that is clearly counter-productive. They become so focused on simplifying some small aspect of the problem that they lose sight that they’ve made everything else worse.

Since I’ve moved around a lot I’ve encountered a great variety of good and insane opinions about software development. I think it would be helpful if we could consolidate the best of the good ones into some single point of reference. A book would be best, but a wiki might serve better. One single point of reference that can be quoted as needed. No doubt there will be some contradictions, but we should be able to categorize the different practices by family and history.

We do have to be concerned that software development is often hostage to what amounts to pop culture these days. New “trendy” ideas get injected, and it often takes time before people realize that they are essentially defective. My favorite example was Hungarian notation, which has hopefully vanished from most work by now. We need to distinguish between established best practices and upcoming ‘popular’ practices. The former have been around for a long time and have earned their respect. The latter may make it to ‘best’ someday, but they’re still so young that it is really hard to tell yet (and I think more of these new practices are deemed ineffective then promoted to ‘best’ status).

What would definitely help in software development is to be able to sit down with management or rogue programmers and be able to stop a wayward discussion early with a statement like “storing all of the fields in the database as text blobs is not considered by X to be a best practice..., so we’re not going to continue doing it that way”. With that ability, we’d at least be able to look at a code base or an existing project and get some idea of conformity. I would not expect everyone to build things the same way, but rather this would show up projects that deviated way too far to the extremes (and because of that are very likely to fail). After decades, I think it’s time to bring more of what we know together into a usable reference.

Sunday, November 4, 2012

Work Smarter...Not Harder

I’ve always loved this quote:

"Work Smarter...Not Harder"
Allan F. Mogensen

But what does it really mean, particularly when it comes to software development?

  • get organized immediately and cleanup often
  • being consistent is more important than being right
  • be very sensitive to scale (bigger systems mean less shortcuts)
  • utilize the full abilities of any technology or tools, but don’t abuse them
  • automate as much as possible, spend the time to get it reliable
  • minimize dependencies, accept them only when there are really no other options
  • read the manual first, even if it is boring
  • research before you write code, avoid flailing at a problem, seek expertise
  • choose to refactor first, before coding
  • delete as much code as possible
  • encapsulate the subproblems away from the system, spend the time to get it right
  • break the problem cleanly, fear special cases
  • apply abstraction and generalization to further reduce the work
  • think very hard about the consequences before diving in
  • fail, but admit it and learn from it
  • don’t be afraid of the domain, learn as much as possible about it
  • focus on the details, quality is all about getting them right
  • accept that complexity breeds below the surface, if you think something is simple then you probably don’t understand it yet
  • know the difference between knowing, assuming and guessing
  • everything matters, nothing should be random in software
  • don’t ignore Murphy’s law
  • a small number of disorganized things doesn’t appear disorganized until it grows larger
  • reassess the organization as things grow, updating it frequently as needed

Working smarter is most often about spending more time to think your way through the problems first, before diving in. Under intense time pressure, people often rush to action. There are times when this is necessary, but there is always a cost. Rack up enough of this technical debt. and then just servicing it becomes the main priority, which only amplifies the time pressure. Thus any gains from swift action are lost and working harder won’t undo the downward spiral.

Being smart, however, can prevent this from occurring. Yes, the pace is slower and getting the details right always requires more effort, but a minimal technical debt. means more resources are available to move forward. Eventually it pays off. Being smart isn’t just about thinking hard, it also requires having enough data to insure that the thinking is accurate. Thus, acquiring more information -- dealing with the details -- is the key to being able to amplify one’s thinking. In software development, what you don’t understand can really harm you and it’s very rare that you can ignore it.

Saturday, October 27, 2012

Setting Up Development

While it is not always true, you’ll frequently find that in restaurants there is a reasonable correlation between the quality of the food and the management of the kitchen. In my youth I had always found that a messy kitchen often meant low quality food and/or bad service. Of course it’s not always that simple, but usually a disaster area behind the curtains percolated out to the rest of the organization. Since I prefer to ‘do things well if I have to do them’ when searching for jobs back then, I stayed far away from disorganized, sloppy or dirty restaurants, even if it was well-hidden behind the scene.

The same is essentially true in software development. Projects that are set up correctly have strong processes, gravitate towards best practices and control the client’s expectations are considerably more likely to produce quality software. The correlation isn’t always as strong though, since I have seen smooth development shops that still produce fugly software. But I’ve never seen a badly organized shop produce excellence.  

One far too common problem I’ve seen is that the client expectations are out of whack. In the printing industry people often say “Fast, Cheap or Quality, pick two” as the means to explain the types of trade-offs necessary when purchasing print work. In software we get the same saying but it ends with “pick one”. I figure it’s worse for our industry because our work is considerably longer. Print jobs are often measured in hours, days or perhaps weeks, while software is usually measured in weeks, months and years. I’ve seen far too many software projects where clients insist on all three and then have a hard time understanding why some “fiddling” with a few screens might require more than a trivial effort.

A development project that is set up well not only attends to the implementation and distribution of software, but it also must control the client’s expectations. They need to be reasonable. Failure to do this means too much pressure on features and quick fixes, which means accumulating a heavy technical debt. Left unchecked, the project spirals downwards until it combusts.

You can see the symptoms of that type of pressure directly in the code base. Poor analysis, little design, poor use of tools and overly frequent ‘micro’ releases. As well, there is generally a high bug count and a lot of time burned on ‘second-line support’. If developers are busy with band-aides then clearly they’re not coding new stuff. Too much support and too little features just amplifies the pressure, which eventually becomes so toxic that big things start to break.

Getting out of this type of spiral can be difficult or impossible. Resetting the expectations of people who are already disappointed is an extremely tough task. However, just like those reality TV shows where they help people in serious financial debt learn about handling their finances (like Til Debt Do U$ Part) in a more reasonable manner, the ticket to really fixing the development problems is to tackle the technical debt, which mostly means being cost conscious (features & make-work) and reducing lifestyle expectations (release dates). Digging out takes a massive amount of time and means a lot less progress, but it only gets worse if it is put off for longer.

Reversing the slide means revisiting the development and release procedures. It also means lots of cleanup and refactoring. And it usually means revisiting all of the development tools to upgrade their usage to known best practices. For confidence, a few low hanging features might be tossed in as well, but all major changes need to be put on hold until the implementation and release process stabilizes. A smooth development effort should produce minimal support effort. It should be reliable and repeatable. The code should be clean and consistent.

Getting back to the kitchen in a messed up restaurant, not only do you need to clean up the work area and replace any defective equipment, but you also need to set some processes in place to insure that it doesn’t happen again. Once that back-room problem has been dealt with, you can consider a new strategy and/or a new menu, but neither of those will help if the kitchen lacks the ability to execute.

You’d think that new projects were less vulnerable to this type of problem, but most often the pressure to get something working right away overrides the desire to set things up properly. The developers start letting little issues slip and then after enough time, they get hit with ‘death by a thousand cuts’. The most visible “source” of the problem is the client expectations, but they only become overwhelming because of a lack of management. Still it is ultimately up to the developers to ‘insist’ that the technical debt not be ignored. At some point it will always be dangerously large, so at very least ‘cleanup’ should be a recurring work item. Personally I like to start each new development iteration with some significant cleanup, mostly because its boring and it gives one a chance to take a breather after the last release. If that becomes a steady habit, while there is always technical debt and time pressure, at least it never builds up to toxic levels.

In the end its not just what you code, but how you code it and how you leverage the development and systems tools as well. Digging out of technical debt is an on-going problem and mostly the fixed time allocated to do the work should be relative to the size of the debt. It doesn’t need to happen in one big effort, but it can’t be ignored indefinitely either. A smooth development process means that the developers can spend more time contemplating the behavior of the code, which is really the only route to quality. A disorganized process means a strong head wind which is only overcome by dangerous short-cuts. Thus smooth is good and it is a primary requirement to achieving quality. You can’t have the later without spending time to get the former.

Monday, October 15, 2012

Programming Style

Does it really matter what the code looks like? The short answer is an emphatic ‘yes’.

It matters because ultimately writing code is about discipline and details. If the code is just slapped together, really messy and there are issues like extra blank lines or extra variables laying all over, it reveals an awful lot about the author. Perhaps they were in a huge rush or even a panic. Maybe they are just lazy or don’t care. Maybe they really don’t understand what they are doing. Any which way, their state-of-mind while doing the work was such that they lacked the self-discipline to clean up their work and thus were probably not particularly good with playing attention to the all-important details either. Messy code is almost always fragile and usually a bug fest.

Code is basically knowledge and intelligence encoded into a sequence of steps for the computer to execute. It is only as smart as the programmer that wrote it, and it is only as flexible as they conceived of the runtime environment. If the programmer’s style if haphazard or their thinking muddy, then the code will suffer the same faults. If however, the programmer’s understanding is crystal clear and they grok the environment, then the code will stand up to a lot and keep on working.

Just getting something to barely work for a extremely limited set of circumstances is not good enough to be considered professional quality. Working code has to work correctly for all circumstances; for all inputs; for all changes in the environment. It needs to be able to stand up to the full range of things that happen in the real world. If it doesn’t, that’s an indication of a lack of quality.

What experienced professional programmers always come to understand is that quality starts in the code. If you want a beautiful system that is a joy for the users, then the code underneath needs to not only be smart and well designed, but it also needs to look marvelous as well. The ‘platinum’ standard of coding comes from taking a complex problem and making the code look simple. Elegance comes from both the readability and the underlying skill used to make the answer look obvious. A well-written piece of code hides the messiness of the details carefully. It misleads a reader into thinking it took just a fraction of the time to get written. It leverages abstraction to avoid redundancies, but not at the cost of becoming obfuscated. It basically takes the knowledge of the programmer and encapsulates that into a small, tight region of the system that works correctly and makes it easy to extend later. It draws clean, hard lines through the problem and then ties them up with the lines in the solution. And because it is easy to see its sole purpose, it makes it really easy to detect and fix bugs. Quality software comes from quality code.

But it is not just quality that is affected by bad code. Testing, bug fixes, and support are all heavy drains on software development. Work spent in these areas is work that is no longer available for new development. Crappy-code bugs are way harder to find and fix, so as a consequence they chew through more resources. It actually costs more to write bad code than doing a good job the first time around, line for line.

And often projects get caught up in a vicious cycle of coding fast and recklessly then loosing too much time dealing with the inevitable problems, causing them to go right back to fast and reckless again. That type of cycle feeds back into morale, and the whole project does a slow spiral down the drain.

As for programming style, mostly it doesn’t really matter which conventions are chosen. Some conventions do make improvements in clarity or readability and help to make inconsistencies more obvious, but so long as all of the code is consistent, it can always be refactored to a better convention at some point.

Getting all of the variable names the same, laying out the lines with the same whitespace formatting and finding strong conventions for naming core parts like methods and objects establishes a strong foundation for the code. Normalization and abstraction help to either hide the details or put them forth in a way that inconsistencies are noticed easily. A solid architecture decomposes the system in a way that makes it easy to trace bugs quickly back to isolated parts of the code base. All of these things help in reducing the number of bugs, and also reducing the time it takes to fix them. And because of them, it opens up more time and resources for new development.

Building a high quality software system always means playing close attention to the millions of details floating about the project. It is not enough to just hack through the work, it is a detail-oriented process, which often needs great time and patience. All development projects come with outrageous demands and too little time, but being sloppy doesn’t help with these issues. Part of the great skill involved in being a professional programmer is the self-discipline to do the right thing, even under intense pressure. Sloppiness and short-cuts, while tempting are just putting off today, what will always cost more in the future. If the goal is quality, then that has to percolate into every tiny detail. Fast and cheap is always crap in software.

Wednesday, October 10, 2012

Knowledge Bases

To me ‘intelligence’ is basically raw human thinking power. It’s our ability to work through problems. In order to employ intelligence effectively in our world, we need data. And that data needs to be structured and interconnected to make it usable to us. Usable, organized data is what I take to be ‘knowledge’. Its all ready to be utilized.

Our various endeavors are categorized into a huge number of fields (or domains) such as law, medicine, finance, physics, math, biology, etc. each of which is mostly an independent ‘base’ of knowledge about the field. What’s really interesting about most knowledge bases is that they are easy to over-simplify and also frequently counter-intuitive in their depths. An outside perspective can easily lead to the wrong conclusions. It’s not until you are steeped in the details, that you’re able to apply your understanding correctly.

Without enough underlying knowledge, intelligence is just hand waving. You can think about something as deep and as long as you want, but your conclusions are unlikely to be reasonable.

Software development is extremely complex. Not only do you need a strong understanding of the underlying technologies and how to use them, but it also cuts across many other knowledge bases.

A software system is a solution to a problem in a user domain. If it doesn’t solve their problems, then it is only adding to them. It consists of a huge number of details, all of which need to be organized and fit into the final work. Managing these pieces, usually under tight time-lines requires a special set of skills. Software is also very expensive to build and maintain, so finding the money to pay for enough resources is always tricky.

Putting this together we see that software spans the following knowledge bases:

  • Technology
  • User domains (all?)
  • Management
  • Business

Most programmers tend towards believing that technology is the most significant issue in software development, but usually the problems start in other areas and feed back into the development work. For this post, I’ll go through all four areas in the order that I find have the most impact for big projects.

Management

A large software project has millions of details that are all being juggled at the same time. Keeping most of those balls in the air, without losing track of them, is necessary to deliver a working system.

A good manager is constantly running around, making sure that all of their resources are moving forward, and that any roadblocks are eliminated as swiftly as possible. Some people see software management as a higher creative input role where they just have to inject ideas, but that’s never actually helpful. Most development projects have more than enough ideas, what they need is organization, process and to keep everyone on the same page. In that sense a good manager accepts their role as herding all of the cats in the same direction, while making sure the path forward is clear.

Development meetings are a good place to assess management. If they are retrospective and basically just a forum for people to catch up with what has already happened, then the project is not well-organized. It shouldn’t be necessary for the manager to catch up, if they have been paying attention as they should.

If, however, meetings are really planning sessions for the next upcoming work, then they contribute by identifying the issues long before they become significant enough to derail the process. Plans don’t always work as expected, but planning is a necessity when you are dealing with long running work. Even small tasks in development can take weeks or months, and the worst possible outcome is to keep shifting directions before any of the work is completed properly. What’s started should be finished, or it shouldn’t have been started in the first place. Only a long-term plan will avoid time-crushing ‘make-work’.

Since development flows through different phases, the management of each has to change as well. A basic development iteration has five parts:

  • Analysis
  • Design
  • Implementation
  • Testing
  • Distribution

Analysis for example is a never-ending effort about depth. Wherever you skimp on analyzing the requirements, it will come back to haunt you as ‘scope creep’. However it would be rare to actually have enough time and patience to fully analyze everything before the work starts. Analysis is also very sensitive to the approach; ask the right questions and you get the right answers. Ask the wrong ones and you get misleading information. Managing analysis means ensuring that both the right questions are being asked, and that the answers are being organized. Experienced analysts need little intervention, but the quality of the work isn’t known until later in the implementation phase, when it is often too late.

Design and Implementation are really about making sure all of the programmers are on the same page; that they are working together well and that they have all of the input necessary to keep them moving forward. For large projects this is basically about team building. A system built by a dysfunctional group is naturally a big ball of mud, and the messiness of the work creates a huge of amount of extra make-work to keep it all from falling apart. A good manager needs to keep the team working, arbitrate disputes and enforce the process and standards even when the time pressure is intense. They also need to protect the programmers from any outside interference to insure that they have enough calmness to be able to think clearly and deeply about their work.

Testing is all about quality vs. time trade-offs. It is impossible to test to 100%, so there is some lower percentage of testing that has to be accepted. Often that might even be just a fixed block of time. The biggest problem during testing is to keep everyone working. Testing is often seen as boring -- particularly after an intense development slog -- and the developers start to lose their attention to the details. Bugs get noticed, but then forgotten. Managing here means dragging people around, holding hands, calming nerves and just trying to get the process as constructive as possible. It can also means having to make brutal choices about quality vs. timeliness. Sometimes the software has to be shipped, even if it is not as good as it should have been.

Distribution of software usually involves an operations dept., or selling the system. Because it is usually forgotten until the end, the distribution is rarely well-planned. Management here needs to insure that the software is supported well enough, but not at the cost of disabling future development work. The handling of feedback and bug reports can all be set up in advance, and most projects require both a standard release and an emergency fast-track process. Most issues take some time, but some need an ‘all hands on deck’ approach in order to deal with them before the situation escalates.

Business

A large software project takes many man-years to build and for most projects the work is never done. It just goes on, year after year. Hiring programmers isn’t cheap and you need a lot of other support staff in the project as well including: project manager, system admin, specialists, etc. Commercial quality systems also need roles like: graphic designer, UX expert, editor, translator, etc.

All together, to build something large that isn’t a weird eclectic mishmash of disorganized functionality means having to shell out a huge amount of money. For software, money is time. That is, if you have enough money, you can hire the necessary resources and experience to get the work done, given some reasonable time frame. A lack of money, means that you have to do more with less, and often times it’s that time pressure that leads people to take ill-advised shortcuts, which will eventually scramble the project so badly it can’t be saved.

As such, setting up a development project always means having to figure out where the money is going to come from. And as is always the case with money, what strings are attached to it. Nothing comes for free.

The business world is very subtle, complex and constantly changing. Most techies who encounter it greatly oversimplify its nature, shades of gray and depth, which generally leads to unrealistic expectations of how things ‘should’ work. Some people have a better intuitive feel than others, but for most people it is best to just write the whole domain off as ‘irrational’ so that they won’t make any false assumptions.

Because it is volatile, the business world inflicts a lot of short-term influence on software development. This conflicts with the long-term nature of the work, so it requires making a large number of very difficult trade-offs. If the work always bends to the short-term concerns it will quickly become a mess. If it always sticks to the long-term, it will likely lose confidence and funding. Thus it needs to perform a very difficult balancing act between the two, so a good leader that finds the right trade-offs will end up taking flak from both sides. If both sides are a little unhappy, then it’s likely balanced properly.

For most people, the ability to balance the business and technical requirements is learned from long, hard and brutal experience. Intuition is usually bias to one side or the other. Even the greatest intelligence can only ‘guess’ its way through the complex interactions, getting more wrong than right. So all that is left is learning from experience; from the successes and mistakes of the past. And it takes some painful introspection to really be objective about the many causes of failure. People like to blame others, but within the spidery web of development, all things are related and no one is immune from influence.

User Domain

Software starts with a problem that needs to be solved. That problem is always domain specific, such as a financial system, inventory or even social networking, etc. Most domains have their own unique set of terminology, usually steeped in history. They often have ‘rules of thumb’ or dirty little secrets lurking in their darkened corners. What they never are, is laid out in a nice rational manner all ready to be modeled in software. Often the data is poorly understood, the process disorganized and each organization within the domain is slightly different. There are always rules, but they are not always followed rigorously, nor particularly logical.

Thus the core problem in software is taking some ill-defined ‘informal system’ and finding a reasonable mapping to a very rigid formal system modeled inside of a computer. In practice this makes the analysis of an existing domain one of the most crucial parts in arriving at a functional software system. However, because it is always gray and messy, it is also the part of the process that people ignore the most often. They just start coding, and hope that somehow, after a while, the answers will come to them. Sometimes that works, but more often getting off on the wrong foot is extremely fatal. If you build too far from the actual answer, then you’ll have no choice but to do it all over again. But since people don’t like to restart their efforts they usually flail at the code, hoping it will somehow work itself out. However if they started really badly, they could bang on it forever and still never get something that works.

A very common problem in development is that the users rarely know what they want, or what would work correctly for their problems. Some have strong opinions, but they often lack a full understanding of the consequences of their choices. Most flip-flop faster than the development can be completed, so tying the process too closely to their wishes frequently results in a mess of half finished, or poorly thought out code. However the converse is also true, code written in an “ivory tower” far away from the users most often oversimplifies the core problems making it somewhat less than helpful. To bridge this gap requires domain experts, and often a very deep understanding of the real problems. Everything the users say is important, but not all of it needs to be taken literally. They usually understand their own informal systems, but the mapping to software, and the code itself are best left to people with plenty of experience. It is easy to write code, but extremely difficult to know what code is right for the solution.

Technology

Last, but not least are the technologies being used. Most projects involve anywhere between 3 and 30 major technologies, and a slew of minor ones. Each technology is eclectic in its own unique way and needs some time and experience to discover its strengths and weaknesses. People often focus on the programming language when assessing skills, but in most projects crafting conditionals, loops and slicing & dicing code are the easy parts. There is skill involved in solving the endless array of coding puzzles, but unless the project is pushing the bleeding edge of technology, it becomes significantly easier as one gains experience. Unfortunately, getting heavily seasoned (>15 yrs) programmers is difficult, since they are expensive and often leave the industry.

Software development is hugely affected by scale and by the desired quality of the final work. For scale I usually break it down as:

  • small - 1 developer, <30 lines="lines" span="span">
  • medium - 1-4 developers, <99 lines="lines" span="span">
  • large - 5-20 developers, <1 lines="lines" span="span">
  • huge - teams of developers, millions of lines

Projects don’t jump scale easily, they often require a nearly complete rewrite to go from one size to another. Disorganization and bad practices often work OK for small and medium projects, but they become fatal beyond that.

As well as scale, the desired quality is important. I generally see it as:

  • prototype - rough proof of concept
  • demo - a small working set of features, mostly works.
  • in-house - eclectic, lots of rough edges, inconsistencies and operational issues
  • commercial -- solid, dependable and beautifully designed by graphic designer / UX experts. Looks good, works correctly.

Each increment in quality takes considerably more work and requires specialists to focus on their particular strengths. Products often sell although they are just in-house quality, but then they are vulnerable to stronger competitors. Users may not complain directly about inconsistencies, but they do generate a negative impression of the system. Badly architected ‘balls of mud’ can often degenerate in quality as programmers randomly slap weak functionality into the corners. Poor development practices tend to be reflected in the interface, so often the outside of the system is a good indication of the state of the code base.

A rather dangerous development trade-off often comes in the choice to build or buy (paid or free). Depending on the technical specs, building is often extremely time-consuming, but in my career I’ve seen more people fail because of their choice to buy. All technologies are a collection of their author’s eccentricities, and often these play a dominate role in their usage. Buying specialty libraries and packages usually works well because you don’t have to acquire the knowledge to build them, but for the core parts of the system,if you depend on someone else’s solution you limit your options going forward. That can drive the architecture and constrain the path forward in dangerous ways.

Getting a commercial grade large or medium system to users is always a team exercise. It takes a large group of people, all with different specialties, to make it happen. As such, the team dynamics become crucial in determining the outcomes. Badly functioning teams usually produce badly functioning systems. Rogue programmers may get a lot of opportunity to express their creativity in their work, but they often do it at the expense of the overall project. Multiple different coding styles and no consistency generates high bug counts and stability problems. Once a project gets going it only continues to work if all of the people involved are on the same page, which generally means strong leadership, a reasonable process and a well-defined set of objectives. Efficiency means long-term goals, even if the short-term is volatile. Initial speed can be gained from brute force hacks, little or no reuse and heavily siloed programmers, but each of these accrues significant technical debt, which eventually bogs the project down into a mess. If the work is getting harder over time that often means disorganization, no architecture, redundancies and/or no long term plan, which if left unchecked will only get worse. A well run, well thought-out development effort will build up a considerable number of reusable common components, which if documented will guide extensions and new features. Architecture sets this commonality and management enforces its usage.

Finally

What sets software apart from most other professions is that it requires a larger cross-section of other knowledge bases to keep it successful. The danger in software is that from an outside perspective, it all looks so easy. You just have to throw together a bunch of simple instructions for the computer and chuck it all onto a server. But that type of over-simplification has always lead to disasters, caused by people who don’t have enough experience to respect the underlying complexity and trade-offs required for successful development. Most other professions are usually managed by people who have moved through the ranks. Software is often special, in that the most experienced developers are rarely put into full leadership positions. More often it is business people or domain experts that attempt to drive the projects forward, although few have the necessary prerequisites. Lack of knowledge usually equates to bad choices, which always mean more work and a much higher likelihood of failure. The complexities, knowledge and work involved in software development are easy to underestimate, so the failure rate is obscenely high.