Opera's Unwanted Functionality

I was just doing a bit of work in the Opera web browser, typing some information into a web app's text box, when I accidentally de-selected the input box in the process of jumping between applications. On my next keystroke the interface suddenly went to an archaic layout. It looked like something rendered in Netscape 3.

I had no idea how I did this, it was completely unwanted, and the impact was extremely disruptive. Closing and restarting the application didn't remove this sticky setting, and randomly (and systematically) selecting what I thought would be the accidental shortcut keys yielded no solution.

Now I had to waste time finding, and then turning off, a feature that I didn't want in the first place.

This brings to mind a couple of user interface issues:

• All applications should have an undo stack (a transaction log, of sorts) that clearly logs and lists every notable app change - minimizing, maximizing, resizing, or changing options like rendering mode. If there was such an undo stack, I could jump to it, see the "Switch to User Stylesheet Rendering" transaction just occurred moments ago - maybe even with a help link to see what that event type was and how I triggered it - and I could roll it back. This is something I've wanted for years (neigh...decades) but of course have never had the time or resources to implement such a non-standard technique in my own apps.
• Having options like this - switching rendering modes - configured for a standard key (in this case Shift-G) is questionable, and given the severity of the change it should really ask for a confirmation of use. "You have selected the shortcut to switch to User stylesheet mode (Shift-G). Are you sure?" (with a checkbox allowing brave users to avoid the confirmation in the future). It's a bit of education and avoidance wrapped up in one.
• Opera doesn't indicate the shortcuts on many of their menu items, including the style mode menu item (the culprits in this case), making it difficult to investigate possible causes.

While there is a minority of users who override site stylesheets with their own, justifying the feature in Opera (though I'm not convinced that it should be an everyday keystroke like Shift-G by default), this brings me to another user interface observation.

Highly Configurable Interfaces are Usually Detrimental

Drawing from personal experience, I worked on a project quite a few years back where one developer insisted upon absolute flexibility in the user interface - Every toolbar had to be movable and dockable anywhere, every sidebar item drag and droppable, every menu item configurable, every UI skinnable. It was a nice cop out for us because we didn't really have to put too much thought into the interface, and could always justify it with the stock "the user can configure it how they want". Stick some more toolbars, statusbars, and panels in there because the user can clean it up according to their own needs, the logic went.

In the field, about 99.9% (more likely 100%) of the time that people discovered this functionality it was to their detriment. Like the taskbar-stuck-perilously-on-the-side-of-the-screen on your Aunt's Windows 95 computer, it was just something that happened by accident, and they didn't know how to get it back the way it was: No one (or very few) did it on purpose, but there it was terrorizing every computer user.

The first step of any support call for our app was to determine in what innovative ways the user managed to mess up their user interface. After getting a visualization of the sidebar on the bottom, the icons all on the background, the toolbar on the right, some critical toolbars hidden, with the menus all jumbled and the icons all removed, the cleanup began.

On the next release a menu item to reset the interface to the initial defaults was added, and on further releases most interface flexibility was removed (or alternately made much more difficult to do - you had to be dedicated and informed if you really wanted to change things. Someone is much more likely to unintentionally hit Shift-G with no input box focused than they are to accidentally go into the advanced preferences and set an option).

The moral of the story is that customizable interfaces are seldom beneficial, and instead function as a lazy, non-committal cop-out by the developers and designers of the application.

Even the most fundamental element of our user experience - windowing - merits some analysis: Apart from Winamp and Media Player, how often are apps in any configuration other than maximized or minimized? I run with dual-monitors, and 99% of the time one or both of them has a full screen application on them. My "windowing" is alt-tabbing through full-screen windows, and I copy data between apps using copy/paste, or, where dragging is necessary, via the taskbar.

Tagged: [Software Development], [Programming], [Software-Development], [Usability]

I've pursued various Microsoft certifications over the years, starting with the MCP, and then acquiring an MCSE and MCDBA.

My motivation in pursuing these certifications was that they served as a destination of sorts, motivating me to learn products and technologies to a breadth and depth that I wouldn't have otherwise.

The knowledge has proven very handy: Even when I serve in a development/design role (especially when I serve a development/design role) the information gained is critical in making appropriate decisions. When I serve leadership and advisory roles, again I'm glad that I spent the time going through every esoteric option and alternative, because the knowledge does help to head off misdirected initiatives and wasted effort.

For those who think "Oh, but I know all of it anyways. I am a Linux super-guru and thus I can achieve anything on the Windows platform with ease". I've heard this sort of boast before, and the results weren't pretty. Go to the Microsoft certification site and take some assessment exams - you might be shocked. The platform is absolutely huge, and it is remarkable how much of it doesn't gain our attention or focus, yet it can help us make better apps, and deploy better solutions.

"If you're a software developer and development manager, why did you get administrative type certifications?" some might ask. Very good question, and the answer is found in the paragraphs above - I dealt with the coding side all day every day, so I didn't see as much of an advantage focusing on an area that I know so well (basically it would have been hundreds of dollars for Microsoft to anoint what I'd proven amply in the field), while I (like most development focused people) didn't really pay enough heed to the platform side of things. Now that I am often called upon for platform consulting as well, it was a nice foundation to build upon.

Nonetheless, now that Microsoft has revamped their certifications, I've decided to upgrade to the MCTS: SQL Server 2005 along with the MCPD. I had hoped to get the MCTS out of the way, but it looks like the exam isn't available yet (despite a November 2005 timeline). Alas. Already it has encouraged me to focus on esoteria of SQL Server 2005 that would have gone ignored.

And for those who protest "But I don't have time! I'm a very busy person!": You could very well be running to stand still. It is an epic problem in this industry that tremendous effort is expended because people don't spend enough time on the skills side of things, focusing all of their attention on the application side.

If you're thinking of providing a demo or limited-use version of your software, pay for the bandwidth and host it yourself. It is an enormous waste of time for potential customers - not to mention that it's incredibly insulting - when you host at one of the big "make you follow 7 links, then sit in a queue, and then download a potentially tampered executable at a reduced speed" 3rd party file hosts that seem to be all the rage these days.

Bandwidth is relatively cheap nowadays, coming in at less than 8 cents a GB at many providers.

What does this have to do with social proof? Well if you host your demo or lite version at one of the aforementioned file hosts, my immediate presumption is that a very tiny percentage of users actually pay for the software: What else could justify such an abuse of clients?

Given this obvious conclusion, the power of social proof pushes me to lean against purchasing it either.

A lot of my work - both system consulting and software architecture/development - relies upon Microsoft technologies: Whether it's re-engineering a legacy system to take advantage of new SQL Server features for performance or functionality, overhauling a network infrastructure to leverage ActiveDirectory and the extensive platform security functionality, or developing a performant and scalable time-tracking application for an enterprise client, Microsoft is often a very important part of the equation.

Partly due to specialization (it's the tools we target), coupled with simply being the best choice in a lot of scenarios in our target market, we heavily rely on the Microsoft platform for ourselves and our customers. As a professional I can say with confidence that the platform is a secure, high performance, extremely scalable, robust one that compares very favourably against all competitors.

That wasn't always an accurate statement, though. Indeed, it is remarkable looking at the history of Microsoft and learning from their success: On paper it really is hard to believe that Microsoft maintained the market dominance that they did, and it's amazing that competitors couldn't capitalize on Microsoft's late entrance into a lot of markets, and their missteps in others.

Was Microsoft a master of timing, holding off on technologies and advances until the perfect time, or were they simply the beneficiary of a captive audience that was willing to wait however long Microsoft took, blind to the available alternatives?

I'll provide a couple of examples that I recall marvelling at as they occurred- these are hardly exhaustive, however I think it's a nice sampling.

Microsoft Maladies

• Microsoft core offerings were crippled by real/virtual mode limits until long after the 386 and 486 were prevalent. In a nutshell, this made software development a lot less pleasant, and the resulting applications more limited and unstable - I remember being enormously unhappy learning real-mode assembly on the x86 after dealing with the elegant, 32-bit flat world of Motorola 68000 assembly. It seemed so primitive that it still existed, or that a software company continued to rely upon it long after it was obsolete and irrelevant in hardware.
• Microsoft's "operating system" for years was simply the DOS command line, and a set of utilities and software interrupt handlers. While Mac users were busy with a rich graphical user interface, we in the DOS world were anxiously awaiting fantastic new features like DELTREE, and maybe a new version of EMM386 to deal with real mode nonsense. It amazes me now to recall actually going to a store and paying real money for a stack of 3 1/2" DOS 5 upgrade disks...6 years after I was programming applications on a richer 4MB platform, here I was excited that himem.sys could free up some of the critical 640KB of low memory.
• Microsoft toyed with windowing systems, finally creating something credible and successful in 1990 (Windows 3.0). In contrast a variety of competitors had fully-integrated, rich, usable, robust Windowing systems many years before - The 1984 Apple Mac being an obvious example, along with the 1985 Atari ST and Amiga...even options on the Commodore 64. I was an Atari ST fanatic in those days, and I marvelled at how primitive the PC world remained even years later.
• It wasn't until Windows 95/NT that memory protection was utilized to avoid processes stomping on each other's memory. Again, many, many years after most competitors had implemented this basic functionality. Instead we dealt with the normal occurence of misbehaving apps taking down the entire system as a fact of life.
• It wasn't until Windows 95/NT that preemptive multitasking was available in Windows. Prior to this a single misbehaving application could capture the CPU's attention and never let it go (never yielding), which was a fairly typical event. The Amiga featured pre-emptive multitasking a decade earlier.
• Microsoft released Windows 95 without a web browser, remarkably enough, finally releasing a barely changed version of the NCSA's Mosaic in the Plus! pack.
• Microsoft 95 was pretty much a security nightmare. Not only was its software far-from-ready to be connected on the public internet - I remember being the unhappy victims of winnuke and friends when I made people unhappy on IRC (you can't please all of the people all of the time), it also had no real file/object security of consequence. While NT was built as a "multi-user" system from a security and kernel perspective, many of the shell and utilities were user unaware, undermining this capability.
• Microsoft's web technologies were far behind the times until Bill Gates' famous speech that changed their direction, reacting to Netscape's lead rather than charting the course. Internet Explorer quickly ramped up and became the dominant web platform - until it became so powerful that the team was disbanded.
• Alternative 3D rendering APIs (Glide and OpenGL) led the way in an area where eventually DirectX would emerge dominant.

I recall during my early courtship with the PC simply marvelling at how incredibly obsolete the platform seemed to be compared to competitors like the Amiga and the Mac introduced years earlier - from graphics capabilities to software to hardware: Everything about it seemed so backwards in comparison to the superior alternatives, yet customers stuck with it. This was the platform that Microsoft wed themselves to, so surely they would suffer as well, right?

Microsoft's insistence on legacy compatibility led to a platform that moved much slower than competitors - Competitors that had the liberty of just tossing it all out and starting from scratch with whatever whizz-bang feature the newest chips offered. Maybe they could run super-stable and super-fast, and offer the developers an elegant platform upon which to perform their magic...but could it run Commander Keen 1 through 3? Could it run that ancient text database app?

Not All Negatives

Of course it's easy to focus on the deficiencies and imagine that they wrote the whole story, but in reality the situation was much more complex. Windows, for instance, pioneered widely-used video card acceleration (I still remember that shiny new Diamond Speedstar 24x. 24-bit graphics, coupled with hardware acceleration of 3D primitives. It was good times running those benchmarks. Of course the Amiga fanatics will point out that it supported hardware acceleration, just as the STe featured a hardware blitter chip, but the interaction between acceleration and the GDI in Windows really set the bar), and Microsoft's push greatly accelerated the adoption of optical media. Windows For Workgroups brought inexpensive networking to a lot of shops (NetBEUI was imperfect, but it was an easy transition to TCP/IP), and Windows in general represented a "good enough" platform for a lot of users. Internet Explorer, for all of its ActiveX "holes" and CSS quirks introduced the rich web model that we rely upon today.

En Route to 64-bit x86

This all comes to mind as the x86-64 transition accelerates: More and more users are starting to switch to 64-bit capable systems, and the 2/4GB limits of our machines is actually becoming a rational limit among desktop users: Everyday users are shouldered against a limit that seemed almost theoretically large just a few short years ago.

Of course Microsoft has been releasing incomplete 64-bit options for years (for instance you could get a 64-bit version of SQL Server 2000 for the Itanium platform, barring a laundry list of exclusions and limitations, and way back with NT 3.1 Microsoft supported 64-bit processors, albeit in 32-bit mode). Now that 64-bit support is finally becoming a critical factor, Microsoft has a wide gauntlet of support ready, and is finally ready to deliver.

Once again when the market really cares, Microsoft is ready. For years some have been talking about the advantage of various operating systems, such as Linux, being availabile on cutting edge processors and 64-bit platforms. For years that has been paraded as an advantage to customers who continued to run their platform on a standard old x86-32 foundation. Yet now that those limits are being reached, and the platform needs to accommodate new levels of capability and performance, Microsoft is ready. Another deficiency overcome.

The Question

Looking at the platform now - the stability, security, and feature set of Windows 2003, a lot of it already existing in XP - it really does seem like a tremendous window of opportunity for the competition has passed: What used to be a crop full of delectable low hanging fruit is now a well protected enclave featuring armed guards.

If competitors couldn't make inroads before, how do they have a chance now? If Linux couldn't capture the desktop market against a monstrosity like Windows Me!, what chance does it really have against XP?

The most obvious answer is web applications - render the operating system irrelevant and you don't really have to compete.

Tagged: [Software Development], [Programming], [Software-Development]

VCRs and PVRs - Small Usability Improvements Yields Huge Usage Changes

How often did you use the scheduling functionality of your VCR to record your favourite television shows, decoupling yourself from the rigorous schedule imposed by the television networks?

The common answer, overwhelmingly, is never. Few bothered using the scheduling functionality, even when it would be beneficial to their quality of life.

This inspired endless jokes about the complexity of "programming the VCR". Even the few brave "wizards" who did bother scheduling recordings generally did so rarely: The hassle of managing tapes, manually setting schedule times, and then having the uncertain-quality result unavailable until completion simply wasn't worthwhile to most people. Many times it didn't work out, and they discovered that they actually recorded 8am instead of 8pm. Whoops!

Even the introduction of Guide+ - a system that allowed you to record a program by punching in a short code - changed the situation little: To many it still wasn't worth the marginal hassle.

The functionality to time shift was there, but few leveraged it.

This topic came up after becoming engaged in an interesting discussion about PVRs versus VCRs, and why the former is inspiring panic and behaviour changes among the television networks, while the latter was largely ignored. Consider that virtually every household in the West had one or more VCRs, yet only a very small percentage have a PVR today (though obviously it's a much greater percentage among the net savvy). Why the concern about functionality we've had for well over a decade?

The reason for the panic, of course, is that the seemingly minor usability and functional improvements of the PVR dramatically increased the usage and utility of the technology: Instead of rummaging for Guide+ numbers in the back of the newspaper, or worse - configuring start and end times manually - one simply pulls up an online listing, selects the programs they want, and selects to record them. The quality is superlative, it takes just a few moments, and they gain the added ability to quickly skip past commercials. Many choose to automatically record every new episode, saving even more time. To put the icing on the cake, there's no hassle dealing with the tapes.

simply reducing the complexity or number of steps marginally can lead to market dominance

There is a valuable lesson to be learned from this: Seemingly minor advances in usability can tremendously alter marketplace success (the VCR was, of course, a great success in marketplace saturation, but that was almost entirely on the merit of playback of pre-recorded content. Few used it to actually record content). Even when it seems like a marketplace need has been functionally satisfied, simply reducing the complexity or number of steps marginally - or reducing the barriers to entry - can lead to market dominance (or market creation). A PVR isn't just a VCR with a hard drive - It completely changed the equation.

Software For Every Need

Consider the software market: By all appearances it looks to be a saturated market - with a solution for every need - but the remarkable thing is that much of it remains completely unused and unadopted. There are countless domains where solutions sit collecting binary dust because the complexity or barrier to entry is too high.

Skype, in contrast, blazed a path of glory and achieved virtual overnight success, yet really it's just yet-another IP voice technology (like we've had since the mid-90s. Sure it added the distributed net, but that's a feature that is a marginal improvement at best). It offered a clear, usable interface, firewall avoidance, and a simple directory for finding the other person, and bam it is getting bought out for $2.6 billion - for doing what had been done by countless competitors in a seemingly commodity market for years before.

FogCreek software has had success simply taking some open source software and putting a pretty face on it, offering a small value-add (avoiding configuring your firewalls) - Making money charging money in a market that people thought was saturated with free alternatives. The web could really be considered a Gopher 2.0, but improved usability enough to be embraced by the everydayman. Bam, the webolution. HTML is absurdly trivial, yet the marginal usability advance of blogs are what made everyone a writer. CSS and JavaScript are both highly accessible technologies, and you can get started quickly by viewing the source of sites you like, again vastly accelerating the transition from initial exposure to actually doing something with it.

What About the Professionals?

Even when targetting highly-trained professionals, immediate "usability" remains critical. Remarkably many of the successful back-end technologies are those that were easy to get started with.

Extraordinary to think that multi-year projects and massive web applications of tremendous scale were built on chosen technologies because they offered a painless, 10-minute getting started setup and tutorial - letting someone start pushing out code immediately - yet in talks with peers I've found that this is frequently the case. Indeed, I will admit to this irrational behaviour myself - several times I considered implementing a project in J2ME (targeting cell phones), but the hassle of setting up a J2ME development platform, and then the pertinent modules for the various phones, served as such a discouragement that I abandoned the project rather than wasting 4 hours dealing with that. In the longer term of a project it's completely irrational, yet it happens.

Of course much of the ASP development community evolved not because ASP was the best platform that was being chosen on merit, but rather because a lot of shops had a Windows NT box sitting there with IIS on it, and they started dropping ASP scripts on it (other languages, like PHP, required additional installations = more trouble). Soon enough these were ASP shops, even though it was almost accidental. Few of them really seriously evaluated the various alternatives.

Of course this was by design: Microsoft, who I spoke about earlier, understands this resistance to learning well. They have entered countless markets with seemingly inferior offerings (at least at first), but because it's there (Microsoft used to rely upon "everything on" by default) and it's easy to use, the marketplace adopts it. SQL Server is a fantastic database system (I personally believe it was one of the best, and is now the best with SQL Server 2005), but a lot of its growth came about largely because it was a trivial install with a simple, ultra-low barrier to entry GUI: Joe Developer installs it from the MSDN discs, prods it for a while, and soon enough he's building the enterprise data system on it. All because it was so accessible and easy to use [Of course many of those database don't use transactions (or they don't properly), and they host terrible schemas, but it got it used]. On the Windows platform a lot of admins did the "install everything" technique, and slowly they sorted it all out and utilized it. This was the way that Microsoft entrenched itself into corporate networks.

Contrast this with other areas where Microsoft hasn't followed this philosophy, and where the results have been much less positive - Even for critical back-end technologies like Biztalk and Sharepoint (both of which yafla provides solutions and consulting for), where you would think it would be soberly analyzed by experts over months of analysis before deployment (and thus requiring significant upfront configuration should be a non-issue), they often see little adoption simply because the install or initial configuration discourages fly-by investigation. Without the initial investigation there is no one to champion it, so it goes unused (despite being fantastic products).

There are countless examples of products whithering because the first install required 40 steps, and then doing the first "hello world" type of project was an enormous hassle. On the flip side a lot of questionable technologies and solutions have permeated largely because it was usable immediately, with little up-front investment.

Minimize Barriers to Entry - Make Your Software Initially Easy

If you make software products, ensure that Getting Started is as painless as possible, and advanced customization options are saved until the user has some experience with the product (literally it should install and configure everything, and start the user off with a Hello World template solution): Even if your customers will need to spend hundreds of man hours specializing it for their needs, they need to see something they can poke at and interact with almost immediately, giving them a sense of accomplishment to motivate them to continue on. 

Once you've gotten the initial time investment, it's much, much easier to require a more involved understanding, and to demand that the user commit themselves to some educational time by the fireplace with the documentation.

We're a very impatient bunch these days, and this is critical if you want success.

If, on the other hand, you're looking at potential markets for software products, examine the usage patterns of PVRs versus VCRs - While the software world might seem full of existing solutions, really the field is wide open for usable solutions. Make an easier to use mousetrap and much of the world will beat a path to your door.

Tagged: [Software Development], [Programming], [Software-Development], [Usability]

One of the huge stars of the early Internet was e-cards - static, animated-GIF, or Flash alternatives to a standard ink and paper Hallmark card, not to mention that you could save yourself a couple of bucks and the hassle of going out and buying and mailing a traditional card (the biggest boon to men like myself was that you could quickly send one off on the day in question - no event pre-planning necessary).

One of the biggest successes of the e-card phenomena was Blue Mountain. My first experience with Blue Mountain was actually by mistake - I was looking for some ski conditions, and instead of http://www.bluemountain.ca/ I entered http://www.bluemountain.com. Of course I capitalized on my mistake, immediately sending some goofy e-cards to friends and relatives. Of course many of the recipients started using the service as well. I suspect that the tipping point for that company - the contagious source vector - came when winter rolled around in Ontario and the powder was just right, and lots of misdirected Ontarians figured they might as well use the service since they were there.

During those early days e-cards were sent in bulk. It was good times in the e-card business, and marketing was easy: Every recipient was encouraged to send their own e-cards, so growth was exponential.

Excite @ Home bought Blue Mountain for a blistering $780 million dollars in 1999.

Of course, as with most quick-ascent fads, people started to question e-cards: It the sentiment was so cheap, and required so little personal effort, was it really a substitute for a real card? Was it just a thoughtless form of spam, quickly discarded and ignored? It didn't take long for that sort of feeling to pervade the marketplace.

If this were an episode of Behind the Music, dark and disturbing music would be playing. The Dire Straights of e-cards were upon the industry.

Excite @ Home unloaded Blue Mountain for $35 million in 2001. If the business were individually valuated today, I highly doubt it would come anywhere close to even that number.

This thought came to mind given the almost-absence of e-cards I've seen this holiday season. Perhaps it's just me, and my network of contacts is unique in its reliance upon traditional cards, but it is a remarkable contrast. I did receive one e-card - from a web host - and quite honestly it bordered on a bit offensive getting a holiday greetings card that I know was spun off from a mass mailing script.

Project Completion

You just finished a gruelling six-month project to create the Greatest Software Widget Ever Made. Your blood, sweat and tears have been invested into what you believe is a first-rate solution, whether solving a corporate need, or building a better mousetrap to bring untold riches to your ISV.

You email news of your accomplishment to peers and/or clients, proudly sitting back to await the certain accolades.

Amidst the expected congratulations, one reply makes you uneasy. "So it basically does what Project X does?" it asks.

Surely they must be mistaken, you think.

You fumble through the feature sheets for Project X to discover that, sure enough, on paper it does everything that your solution does, and in some ways it is arguably superior: Maybe it supports a richer set of features, is cross platform where yours isn't, or it's a component of a more-complete integrated solution where the whole is greater than the parts.

On paper, at least, it is a better solution.

The Damage

If you're an ISV, you've just discovered that you have more competition than you thought you did. This could impact your market in a wide range of ways.

• Perhaps it's a very expensive product from a company like BEA. You can easily underprice them, and it represents little threat to your product (it might actually help you make sales, giving you a much more expensive product to contrast with in sales pitches). Great stuff!
• Perhaps it's the complex potential use of an existing system that requires such a high level of expensive modules, customizing and programming that it turns out to be much more expensive and risky to actually use.
• Maybe it's an esoteric open-source project with a highly convoluted code-base, few developers behind it, and it requires significant knowledge to configure and use. It represents only a theoretical risk.
• In the worst-case scenario, the alternative is a free, comprehensive, elegant and usable alternative that is quickly becoming widely known. Your market has been devastated, outside of selling in hopes that your customers are blissfully unaware (which is rightfully morally difficult for most people).

If you're a corporate developer, it could simply be a demotivating signal of time wasted (developers are heavily motivated by seeing their solutions doing "good" in the world. Dumping code they invested time and passion into is heart-wrenchingly demotivating. "What have you been working on?" "Well, I spent the last year working on a project that was dumped for Microsoft BizPoint 2009 before delivery"); it could just be ammo for the office Machiavellian (I've was criticized in the past by just such a coworker for not leveraging a "possible solution" that wasn't even invented until two years after my project was deployed, so such Machiavellian Jr's aren't even bound by the constraints of time); or it could be a career killer.

In the corporate space things can get even uglier, as products from the big names like Microsoft will be preferred, however ill-equipped for the task they actually are: If they marginally seem to apply to the problem, there will be those who push for it. Sometimes this strategically makes sense (e.g. if you adopt a new product early, it'll likely evolve quickly if it's developed by a company with the resources that Microsoft has, for instance), but often it's just corporate hubris. Some organizations have a love-hate (mostly hate) relationship with their IT department, so any opportunity to undermine and reduce the dependency on them, even if it involves a more expensive, less-flexible, less-empowering solution, will be pursued. I worked at one shop where a large team of developers worked for years perfecting and expanding in-house solutions, while at the same time the organization continually and openly quested to replace everything they did and were doing with externally sourced solutions. The effect on motivation was profoundly negative, but it was the result of a Better Not Be Invented Here mentality, coupled with a negativity opinion of IT.

The Smaller Scale

Of course such an overlap of functionality or use isn't limited to whole products. It applies to even the smallest component of our projects.

Developers face this sort of issue regularly: From cryptography libraries to database abstraction tools to MVC frameworks to simple coding standards - For virtually every single need, there are solutions that purport to solve all ills and fulfill all needs. Somewhere amidst the 100,000+ version 0.1 projects on sourceforge, the huge catalog of software and feature lists provided by organizations like Microsoft and Oracle, or the millions of projects on the web, and even amongst the libraries and frameworks already included with our platform, there are possible competitors.

From a more optimistic perspective, they are possible allies - code and existing research and knowledge that you can leverage to make your solution better. Why waste time building a dubious non-core cryptography function just to enable security, if you can just use the System.Security.Cryptography collection of classes? Why re-invent the wheel when a lot of standard solutions are crystallized in industry patterns?

The problem is that you have to know about the knowledge, competitors and/or allies that you can leverage.

The Problem

Several months back I wrote that internal code reuse can be dangerous, the premise being that most teams don't spend enough time investigating what is already available (often available in a superior, more widely-known fashion) - be it coding standards, frameworks, or whole projects, a lot of the code base in many organizations is completely redundant, inferior duplications of what already exists, created purely out of lack of awareness. Even when resources are within grasp they're often ignored - A huge number of in-house .NET "utility libraries" duplicate what exists in the .NET Framework. All because no one bothered looking through the admittingly-massive MSDN Library before undertaking a task themselves.

While sometimes this occurs because of the Not Invented Here syndrome, more frequently it's simply ignorance - Either intentional ignorance (such as not bothering doing proper competitor/options research before getting started), or unintentional ignorance (there is such a huge number of projects out there, often with convoluted explanations of what they do, and difficult and time consuming setups to test it out yourself, that it can be impossible to know what already exists. For some small components and projects it can literally be faster to just do it than it is to investigate the alternatives).

Huge code repositories grow larger with second-rate solutions, products are developed that enter the market dead-on-arrival, and developers have the passion beaten out of them.

The Paralysis

As bad as all of this sounds - making solutions to discover that they're second rate or obsolete, or huge internal code repositories filled with duplications of better alternatives - the problem exists even where these problems aren't evident (in fact a secondary-effect occurs because of the prevention of these issues): Many software developers are simply decision-paralyzed. They're unable to proceed with any project because the number of possible standards, components, and frameworks that they need to consider is unending.

This is made worse by the fact that there are tens of thousands of software products and components making claims that they can't back up (especially in realm of nebulous, all-empassing applications. These high level applications often require tremendous configuration and programming to be usable, but the appearance is that they're a drop-in-place solution). Sourceforge, for instance, is full of tens of thousands of version 0.1 applications, where the purported feature list could be be described as a wish list. Other products entail tremendous expense, or bring along undesired dependencies or complexity.

Researching, and then understanding every caveat and issue with every potential solution is a huge undertaking.

The Solution

If only it were so easy to simply rattle off a solution.

Nonetheless, a huge step in the right direction would be simpy allocating adequate time upfront for "competitor research" before committing to a project, or even before undertaking a component. This is a tremendous undertaking these days, yet in timelines it's often forgotten or considered satisfied by a quick 5 minute Google search.

If you're developing a web application to store documents, could it be satisfied by Sharepoint? Community Server? Zope? A derivative of Media Wiki? If you've committed to one of those projects, but you just need to extend some of the functionality (say you want to display the EXIF data of photos), have you committed the time to investigate the alternatives?

For virtually every need, there are a slew of theoretical solutions. The truth, however, is that many of those solutions aren't a good fit - many of them are simply terrible - but you need to be informed and prepared when peers, coworkers and clients ask about them in comparison.

Even if you don't find an appropriate solution already available (in most cases you won't), document all of your findings: If Biztalk looks superficially like it's the right solution, but for some reason it isn't appropriate, be prepared for the inevitable grilling you'll get every single time someone who knows about Biztalk and then hears about your project gets within earshot.

Be prepared!

Tagged: [Software Development], [Programming], [Software-Development]