Thread by @berndschiffer

Australia's welfare program @Centrelink has an average phone wait time of 15 min. They are currently doing a 7-year project to upgrade their IT which costs "billions of dollars".

Seven. Years. For a single project. For billions of dollars.

Oh, it gets much worse.

-> Thread

They are already 2 years into this project. At this point, they can't give any indication about by how much they are going to reduce the average wait time in the end.

The very thing they set out to achieve is still an unknown. After 2 years. After more than 1/4 of the overall project length.

Source: Video "Please hold for Centrelink" by @theprojecttv, 23/01/2018 tenplay.com.au/channel-ten/th…

See the interview with @HankJongen, @Centrelink general manager and spokesperson, about 2 min into the video.

I have questions.

1) Why, after 2 years into the project, isn't there already a reduction in wait time and therefore a benefit for the users? All benefit only at the end of the project?

2) With 2 years into the project, why is @Centrelink not able to give a range of expected reduction of wait time over the coming years? No learning within 2 years?

3) Why is it possible to spent "billions of dollars" on a single project without a regular show of progress? Why so much risk?

4) How likely is it that this particular 7-year project will finish on time and/or on budget? Why is there no learning from other big (and failed) projects, e.g., Myki?

5) How likely is it that there will be a reduction of wait time at all within or right after the project's lifespan? Like flying blind for hours and still landing on the right airport.

6) If there is no specification of a target wait time nor any show of progress, how can the success of this project be evaluated during its lifespan or at the end? How could it possibly ever fail?

7) Why did the government wait that long to upgrade a 30 year (!) old system?What are the costs just because of this neglect?

With only these bits of information, it sounds to me as the riskiest project approach I've ever heard of. By far. Maybe just a very bad interview?

A few more interesting tidbits about the @Centrelink system and upgrade:

It's a "30-year-old system consisting of 30 million lines of code and undertaking more than 50 million daily transactions is responsible for delivering around $100 billion in payments to 7.3 million people every year" ( itnews.com.au/news/cabinet-a… )

The business case for the upgrade was created in two years (2013-2015). The upgrade started in 2015. It was estimated to cost $1-1.5 billion (contract was $1.5 billion). Benefits were anticipated for 2016. ( itnews.com.au/news/cabinet-a… )

Why is @HankJongen talking about "billions of dollars" in early 2018 when the contract volume was $1.5 billion in 2015? The project ran already out of budget, a budget for 7 years? Couldn't find anything about that online.

When (if) they are finished with the upgrade in 2022, the upgrade will be based on up to 9 year old specifications. Seems like they are working to create the next (30 year old) legacy system.

"…up to 36 specialist IT teams can be needed to make changes and test the system. Updating a phone number on a letter can take three months and cost $20,000. Changing a letter … can involve 100 public servants six months and cost of $500,000." ( businessinsider.com.au/heres-why-it-c… )

For a system like this, there must be hundreds of change requests per year. Why is there not an ongoing effort (and budget) to upgrade the system continuously? It would finance itself by avoiding costly changes in ancient legacy systems.

Who's doing the @Centrelink upgrade? Capgemini, Accenture, IBM, HP, and SAP. ( itnews.com.au/news/ibm-hp-ro… )

I'm done here for now. Learned a lot tonight. Still, lots of questions left. Thanks for staying with me until the end of this (loooong) thread here.