If you were a teenager when plans for a new MRT signalling system were first announced, you would be well into your 30s before you see any meaningful outcome of that project.

In July 1997, then SMRT managing director Kwek Siew Jin announced that the North-South, East-West lines would be upgraded. The project would have cost $100 million, be completed by 2002, and would have allowed trains to travel at a peak frequency of once every 90 seconds. That project never began.

Fast forward to 20 years later, the current resignalling project costs nearly twice as much, is slated to be completed by next year, and will allow trains to travel at a peak frequency of once every 100 seconds.

The full benefit of the long-delayed upgrading, which got off the ground when the contract was awarded in 2012, will only be felt in 2019, when the full complement of new trains arrive and are deployed.

If Singapore had carried out what Mr Kwek had announced 20 years ago, many of the woes commuters have been facing in recent years - and indeed, will continue to face for some time - would have been less severe.

SMRT, back then, had already recognised that the system was soon about to reach design capacity, and steps had to be taken to increase that capacity.

Then, MRT ridership was below a million a day - less than a third of what it is today.

But, for reasons that are still unclear, the project never took off. It was talked about time and again during the 10-year period when Ms Saw Phaik Hwa was at the helm, but nothing gained traction.

It was only after the major twin breakdowns of 2011 - in which the North-South Line shut down on two days - that plans to rejuvenate the two old lines were set in motion. It is regrettable.

But is there a point in crying over spilt milk? Is it useful to talk about water under the bridge?

Yes. Because there are valuable lessons to be learnt from the past. And if we fail, or refuse, to acknowledge this epic failure, we are doomed to repeat it.

The task of bedding down the newly-installed signalling system, and the ensuing breakdowns and delays in recent months attest to the importance of upgrading ahead of time.

If resignalling had taken place as originally planned, the disruptions of such a project would not have had such a great adverse impact as they do today.

If nothing else, the sheer volume of rail ridership - at well over three million trips a day - magnifies the smallest of technical glitches.

Our rail problems are not unique. Delays on New York's iconic subway have more than doubled in the last five years (a delay on the 112-year-old metro is defined as any train reaching a terminus station five minutes behind schedule).

Most of the delays were caused by overcrowding, when trains needed to stay at stations longer than usual because more people wanted to get on (or off).

Ridership on the subway has grown by 50 per cent in the last 20 years to six million a day.

But in New York's case, and in the case of metros in many cities, the inability to foresee an approaching crisis is only one part of the problem. The other part is not having the financial resources to fix it.

Singapore's legacy of prudent fiscal policies and a sound financial framework have ensured that it is rarely short of funds for public infrastructural projects. So, all it needs is to know when to spend the money.

The lesson from Mr Kwek's derailed plans must surely be that rail capacity is a moving target - a fast-moving target.

Policymakers need to project what the demand will be in five, 10, 20 years down the road. But first, they must know what the design capacity of the current network is.

Just like airplanes and airports, you cannot increase capacity overnight by buying assets off the shelf. Infrastructure, too, needs time to build and renew.

In the case of the North-South, East-West lines - which remain the backbone of our rail system - there have been ample signs that things were awry well before the two pre-Christmas breakdowns of 2011.

Commuter complaints about overcrowding were becoming increasingly frequent. Glitches and delays were becoming more common. The Land Transport Authority's (LTA's) own statistics of disruptions showed as much.

These events took place between 2004 and 2010 - when Singapore's population grew sharply - and were widely reported in the press. Yet, there was general inaction.

We must ensure this is never repeated.

Since 2011, the Government and SMRT have been spending hundreds of millions on renewing the older lines. And the country has spent billions more on building new lines.

But will it be enough? First, as mentioned, capacity is a moving target. By the time all the additional trains arrive in 2019 for us to reap the benefits of the North-South, East-West lines' resignalling project, rail demand would have grown further.

Second, a competent rail service is not just about hardware. It depends very much on software too.

SMRT is now struggling to restore the system's reliability. From an engineering standpoint, it is doing all that is possible.

But customer care goes beyond that. Even though the operator has a battalion of customer-service officers - who do a creditable job at the stations - there is evidence that these men and women on the ground are not always armed with timely information from above during rail delays or disruptions. That is clear from the recent signalling glitches.

SMRT and indeed the LTA need to know that transparency and clarity are vital, especially when faults arise. Station and train announcements, when there are delays, should be openly and clearly disseminated.

Likewise, alerts on social media platforms have to be put out as consistently as they used to be.

In the recent delays arising from the teething issues with the new signalling system, SMRT and the LTA have asked repeatedly for the public's understanding and patience.

It is a reasonable request. But the operator and regulator must first help the public by releasing information that will help them decide whether to wait out a glitch or to make alternative plans.

It has been fairly consistent in that respect after 2011, sending quick Twitter alerts when flaws exceeding 10 minutes arise. But in recent months, one wonders if lessons from the past are fading.