First of all, let us debunk the idea that Open Workbench is truly open source. Ok they have a website finishing in '.org' suggesting a non-commercial endeavour but when we look at the birth of Open Workbench we soon realise that the whole thing is a bit of a subterfuge from Computer Associates (CA). A few years ago, CA were competing commercially with the other big guys like MS Project, Primavera and the rest of them, when they realised that not only had they become truly out-classed but also seriously out-spent. In the commercial world, Project Workbench - as it was then known - was moribund; in fact pretty much terminal. So in a last desperate bid to get back into the race they decided to simply give it away; thus was born Open Workbench... a gift to the community.

To be fair to CA, their 'sponsorship' of the Open Workbench website is evident and it does not take a genius to realise the collusion with CA’s excellent (paid-for) Clarity product. Yes, read this sentence again, I said CA’s excellent Clarity product! Absolutely, this article is not about bashing CA. In fact, a little bit later, we will come back to Clarity and say some more nice things about it.

For now however, let us look a little more closely at Open Workbench. On the face of it, and if you believe the website, Open Workbench is a worthy competitor to Microsoft Project. Clearly, MS Project is directly in its line of sight. The Open Workbench home page even features a counter which purports to show how much money is being saved  by not purchasing MS Project and downloading Workbench instead. And at the level of 'feature marketing' Workbench certainly appears to measures up. All familiar functionality that a professional Project Manager would expect seems available: Resource Scheduling, Plan Scheduling, Inter-Project Dependencies, Baselines, Shared Views, Calendars, Loading Patterns/Contours etc... But the thing you will most likely notice when using Workbench for the first time is that it has frankly not quite matured into a fully-fledged Windows product - the word clunky springs to mind. Annoyances are rife: inability to use the familiar right-clicking context menu, inability to easily insert additional columns, convoluted way to save specific views, inability to automatically remember the workspace from one session to the next, complicated steps required to perform a mere textual search on a list of activities, inability to split a single task into multiple recurring periods, no useful tool-tips to highlight scheduling conflicts just meaningless coloured cells– and the list goes on – it is painful.

This product was clearly architected in the pre-Windows era (yes a while back) and was never fully refactored to take advantage of what most Windows users take for granted. Fine, I hear you say, it is a little quirky and probably not as user-friendly but once you get over the hump it is still a very powerful piece of Project Management software. After all, there is always a learning curve with new software, isn’t there?

Fair point, but that’s not it. The fundamental weakness of Open Workbench is in fact at the heart of the scheduling engine, in a feature called ‘auto-schedule’. Here is how it works: you enter a list of activities in Workbench and choose to make some of them dependent on the completion of others – basic Finish-to-Start dependency; the most common type. The terrible thing is that Workbench will not automatically honour these dependencies. That’s right, a task's start date will not change to reflect the completion date of its predecessor(s), not until, that is, you run auto-schedule on the plan... the whole plan – no choice about it! The problem is that auto-schedule tries to be very clever in mysterious and intricate ways and will make all sorts of decisions about all your dates, not just task dependencies. BE WARNED. 

Now let us look at the real world Project Managers operate in. A typical project plan will probably contain 1,500 to 3,000 lines. The plan will constantly change and be highly complex, dependencies will be critical and need to be represented, but many activity dates will not be explicitly constrained. The work involved in constraining every start or finish date (“do not start before” etc) is often simply beyond the achievable. Yet if you run auto-schedule on a typically unconstrained plan like this, every date that was not explicitly pinned will be moved. Future tasks not started will be brought back to start immediately, because there was no constraint to prevent that. Likewise, tasks that should have started but didn’t, will be brought forward from the past to start immediately too etc... You can imagine the effect of running such indiscriminate scheduling on a whole plan of 2,000 lines. And good luck to get it back to where you thought it was. System designers will argue no doubt for the theoretical ‘correctness’ of the Open Workbench model. Surely if a task that should have started has not, what’s wrong with moving the start date to the current date? The point is that you don’t want your software to be too clever. With large numbers of tasks it is much easier to see which are genuinely late if they are left where they should have started rather than be blindly moved to the current date.

MS Project does not entirely have a clean sheet on this account, because it too has a feature called ‘auto-level’ which performs a similar job. But the fundamental difference is that with MS Project you do not need ever to run auto-level to benefit from automated dependency scheduling. In fact most Project Managers I know, steer well clear of auto-level for the very reasons outlined above – it’s a little wild and too clever. However, the beauty of MS Project is that if I change the end date of a task with successors, then this will be automatically reflected in the start date of the dependent tasks, whilst the rest of the plan will be left well alone. And that’s what a Project Manager wants.

With Open Workbench, you have no choice – if you want your plan to reflect your dependencies you have to run auto-schedule. Your only alternative is complete manual management of your plan – not feasible in the real world. It’s all or nothing. Thankfully MS Project has struck the correct balance of automated intelligence and manual management, whereas Open Workbench has failed to cater for the real world, opting instead to fulfil a theoretical scheduling model, which bears little regard to how Project Managers actually work.

Unfortunately for Open Workbench, failing to deliver a sensible scheduling engine should pretty much kill it stone dead, regardless of what the website says. So how is it that it survives to fight another day and even spread? The answer I suspect is a combination of sloppy or superficial product assessment by ignorant procurement teams and pressure from the accountants to bring down the very visible costs of purchasing software. MS Project is not cheap and neither are the corporate licensing options from Microsoft straightforward. The idea that you could procure an equivalent product totally free of any charge is certainly an appealing one. But there is perhaps another reason. CA are also the makers of Clarity. Clarity can be fully integrated with Open Workbench to provide a really powerful enterprise Programme Management tool. Clarity is delivered as a website (usually through an intranet), so deployment is simple and it allows Project Managers to access a central tool to manage their risks and issues, their resources and time booking, their budgets and even reporting. Clarity is well-designed, flexible, intuitive and requires very little time investment to climb the learning curve for Project Managers. Clarity however is not free. Again Microsoft have a competing offering in Project Server, but Clarity is a mature and well-designed product, which unlike Workbench can stand on its own two feet.

Sadly, in the environments where Clarity has been implemented, and therefore Open Workbench adopted by default, I have witnessed an illicit corporate counter-culture of Project Managers who use their own equipment and software (read MS Project) to produce plans and use the ubiquitous ‘cut and paste’ to get plans into the requisite format in Open Workbench. Others chose to manage their plans in the less-than-adequate Excel to try to avoid fighting the weaknesses and clumsy interface of Open Workbench. More generally, plans tend to be of inferior quality and the consequences for project delivery can be measured in more than staff frustration. A poorly managed plan on a large project can lead to costly mistakes. It is perhaps time that organisations using Open Workbench took another closer look at the real cost of using an inherently weak tool to manage major business investments in project delivery.

The Open Workbench website can be found at http://www.openworkbench.org