Where did Udder come from ?

Michael Larcombe commenced work on Udder as part of his Ph.D. thesis in 1986. He graduated from Melbourne University in 1983 and worked in Veterinary practice in Maffra (Eastern Victoria, Australia) for three years. The Ph.D. was the escape route from lame cows and pregnancy diagnosis!

His thesis entitled "The economic effects of manipulating reproduction in dairy herds that graze pasture" was completed in 1989, and Udder was born. Udder was used in the thesis to determine the financial impact of changing the calving pattern of a dairy herd.

Little did Mike know when he commenced his thesis, that the answer to the question "What happens if I change the calving date of my herd?" will depend on the management you impose in response to the change. To answer the question properly, you had to predict the outcome of changing the management of the whole farm.

Udder Version 1 was born on a 286 NEC Powermate computer, powered up with a 10 mbyte hard disc and 640 kbytes of RAM. It was a slick machine ! Only took 50 seconds to complete a simulation.

Version 2 quickly followed once Mike started work as a consultant with Maffra Herd Improvement (www.mhi.asn.au), where he is still employed today. This was when Udder began work in the real world. It quickly became clear that it would have to manage predictions where cows were grazed off the farm and a heap of other management options that had not been included in the initial version.

In the early 90's, Udder moved to New Zealand. The early adopters came from consultants working in the newly commercialised Ag New Zealand. One of the earliest adopters was Neil McLean who still supports Udder in New Zealand today through his consultancy business "Agricultural Business Associates".

The work of Neil and his counterparts in NZ resulted in new predictive equations being developed that were robust under most grazing conditions on both sides of the Tasman.

By this time, computers had sped up a bit. Simulations were down to 5 to 10 seconds (10 per minute).

Users were no longer happy asking "What happens if I ....?". They were sick of having to think of new ideas to try. They wanted optimisation to tell them "The best management system to try is....".

Optimisation began in 1992. It only took 40 minutes to an hour to complete one (400-1000 simulations), and we used to queue the computer up to work all night to try out half a dozen option.

Mike soon found that farmers would look at an optimum solution, but would then say, "but what would I lose if I only increased stocking rate to 3.4 rather than 3.7 cows/ha".

It took a few years for computers to catch up, but eventually factorial stocking rate experiments appeared that allowed response curves to be calculated. That is, the best management system and gross margin achievable under a range of stocking rate and milk price combinations. These were pretty impressive processing tasks. They involved 18 optimisations overnight.

Udder went through eight versions in DOS before finally launching into the Windows environment in 1999. This again improved processing performance and dramatically improved the interface and graphics.

Today, Udder's optimiser takes a couple of minutes to run, and it heralds the next major advance. Real time optimisations that will allow superior management systems to be calculated in minutes. That has enabled Udder to become self installing for a farm, and enable farmers to use the product directly without the need for detailed training on how to establish a model for their own farms.

Given the journey we have seen, where will it go in the next 10 years ?