October 4, 2014


Tags: Memories, Work

After leaving Beacon and finishing the MBA, I jumped on a flight to London in mid 2002. Perth was too constraining, I didn’t want to do consulting or mining software for the rest of my career. London seemed like the land of work opportunity. It would also be easier to travel around the world from a base in Europe. Both these reasons turned out to be correct, but I forgot the downsides of London - the expense, the crowds and strangely even the weather. For a good five years or so the positives outweighed the negatives, but then the balance shifted. However, upon arrival I was extremely optimistic.

Crashing in the spare rooms and sofas of old uni friends (the Perth to London road was well travelled by my age cohort), I started to learn about the local IT job market. 2002 was the middle of the dot crash slump. Stories of easy jobs with large pay rates and regular job hopping abounded, but they always ended with “of course that was before the crash.” Most of my acquaintances said that contract work was the way to go - better pay and less bureaucracy, at the expense of career progression (none of my friends planned to stay long, so that was not an issue). I aimed for the same, but in the downturn a person with no record in London couldn’t get through the door for contracting roles - I was too much of an unknown considering the large number of people looking for work. After a few months I switched to permanent roles and started getting interviews again. Lastminute was a notable interview because the interviewer was so offensive I pledged to never use their service - a promise I hold to this day, despite being right in their target market. After 4 months my funds were evaporating quickly and if I didn’t get one of the two jobs for which I was interviewing then a return to Perth would be required. RBS was my preferred option (and I even delayed the other possibility while waiting for them to decide). However, RBS decided I was “too cold” for a management position (at least they gave a reason), so I went to work at the MCPS-PRS Alliance.

PRS Logo

The Alliance is the UK company that controls music copyrights. It is sort of like a club - songwriters register their rights with the Alliance. Then it collects any royalties on their behalf and distributes them (after deducting costs). As the effort in dealing with record companies, radio/TV stations and other business using music is non-trivial, it makes sense to spread the costs across as large a number of artists as possible - that way they all earn more from their royalties. Thus the Alliance is the only firm doing this in the UK (as it has done since the start of copyright law in the country, when it mainly collected royalties for sheet music. It has the scale to do the job efficiently (others have tried and failed). The Alliance is split into two parts, MCPS handled the physical rights (for example, the royalties due to the songwriter when a CD is made); and PRS which handled performing rights (the royalties due if a song was played on the radio or TV).

I was on the PRS side and this is how their business works. Songwriters register their songs once written detailing the fractional ownership. This often goes down to 1/128th of the song and can get quite complicated. Firstly recording companies normally take a large cut for producing the song, but legally they can not take more than 50%. After that songs are often written by teams, so they need to decide the split between them. Then there are cover versions and samples, all of which complicate the situation. Take for instance the Gipsy Kings version of My Way - the Gipsy Kings get some of the royalties for translating the lyrics from the English language version (made famous by Frank Sinatra) and rearranging the music, the person who wrote the English lyrics gets some ownership, then the songwriter and lyricists of the original french version are also listed. Of course all the recording companies down the line also take a share. There is also the humorous case involving John Cage’s 4’33 (a song consisting solely of 4 and a half minutes of silence). Another band create a song of a minute’s silence and called it a cover of 4’33”, Cage sued for copyright infringement and won. Getting the ownership correct can get quite complicated.

PRS has a database of ownership records for every UK song, it doesn’t matter if they have never actually been performed, it only matters that a member has registered them. They also have ownership records for any foreign song which is likely to be played in the UK and is controlled by a friendly overseas copyright authority (which is most of them). This corresponds to millions and millions of records, many of which are very similar. Again, think of all the cover versions of My Way, each with slightly different royalties. The database had 100’s gigabytes of data.

Broadcasters send in lists of the songs they played. Lists come in from radio stations, TV stations, concerts, jukeboxes, and there are also surveys from companies that play music in the course of business - like muzak or hairdressers. Think how many songs a radio station plays in a day, now multiply that by the number of stations in the UK. Similar for TV stations - remember that the jingle in an ad or the intro music to a show all count and need to be listed. At least tens of thousands of listings come in every day, some days it numbers in the hundreds of thousands. Also, the quality of this data varies enormously. The BBC sends nicely formatted electronic records of song, artist, and duration played every day. A concert hall may send handwritten set lists once a month or less.

All the listings go into a batch system that attempts to match the songs played to royalty records in the database. Most of these matches could be done by computer, but many need a human to work it out correctly. There are a couple of rooms of people manually matching songs. Money would then be allocated on a percentage basis from the fees paid by the organisation that played the music. So if a radio station paid £1000/month and one month they played a song 1% of the time, then that song’s owners would share £10 between them, split according to the royalty records. This is totaled and paid out every quarter as long as a threshold amount is reached (£10, I think). The majority of artists in the system don’t earn enough royalties in a quarter to receive a payout. The system is hugely hit based and skewed towards a few big earners. Although a surprising number of people can make a living from it. Apparently the people who write ad jingles do quite well.

By the time I arrived the system was beginning to come under strain. It could still process a busy days worth of listings in a day, but it was getting close. At the expected rate of growth the system would need improving or replacing soon’ish. However, the existing system was originally completed in the early 80’s and ran on mainframes (some of the people from that project were still around in the early 00’s when I was there). Software and hardware was being end-of-life’d by vendors. A complete replacement would be required. As the Alliance did not have the skills for a large new development in-house, it was decided to contract a large US-European IT consulting firm. It was also decided that the new system should be web-based, written in Java and use the RUP. The Alliance had no experience in any of these areas, while I had experience of all three. My job was three-fold: keep an eye on the consultants and check they were acting in the interests of the Alliance; train up the Alliance’s in-house people in the new technologies being used; and, work as a team member and manager in developing the new system.

At first I worked with a couple of the consultants on a prototype while giving presentations on the RUP process to managers and executives at the Alliance. Soon a small team of Alliance developers needed to be trained in Java. Two days a week I was at the Alliance offices in the West End of London. The Alliance people were all very friendly and social - teams tended to eat lunch together, which in my experience suggests good project dynamics. All the teams had favourite pubs and there would always be someone around you knew out on Thursday and Friday nights (usually other nights too). It seemed like a very comfortable and easy place to be. Many people had worked for the Alliance for decades. The company also took training more seriously than anywhere else I’ve been. The Java developers I mentored were internal IT people being cross-trained in Java from scratch (including nearly a month of external courses). I went on two training courses in the short time I was there, including a trip to Darmstadt in Germany. Also, the managers were good and actually managed people.

The other three work days each week were spent at the consultants fancy new office south of London. This was not such a happy time. Although it was a much nicer commute, as most people went in the other direction. Initially, the consulting team was four people in total: two developers, manager and me. The consulting team was a little cold and we didn’t become friends. In fact there were a number of minor clashes. I never felt welcome and this clouded my experience. The team had enough talent to get the prototype done successfully and only a little late. The other devs were certainly good enough at their job - solid programmers who had done this before. Although it was here one of the managers performed one of the stupider planning tricks I’ve witnessed.

Once the project started properly the problems began. Firstly the consulting team greatly expanded, and the quality of the consulting developers greatly dropped. The standard bait-and-switch I’ve seen consulting companies perform a number of times. Also, more of the business people who would actually use the system began to get their hands on the work-in-progress. Many changes were requested and timescales began to slip as the requests were processed into the plan. The basic problem was that a web-based system was not suitable for the heavy data entry the system’s users performed. While the computer matching would be better than before, the human based matching would be much slower, so overall the system would be about the same as the old version. The supposed benefits of a web based system turned out to be spurious - thin clients and working from home were not important to the business users, but speed of data entry was crucial. In retrospect this is obvious. I consider it a personal failing that I didn’t see that the choice of technology was wrong. Although in my defence the technology was chosen before I was hired and was actually the reason for my employment. I took the choice as axiomatic and didn’t question it.

The project problems led to one of the funnier moments in my career. One day I was the only Alliance person at the consultant’s office when admin assistants moved through the area and told everyone to assemble for an all-hands meeting. A new project manager had just started as the previous acted as a sacrificial lamb for the increasingly obvious issues. The new manager introduced himself with a speech to the whole consulting team, and me. He said that after reviewing the project he felt that the the team was working well and the problem was substantially the fault of difficult customers. Then came the punchline, the new managers proudly boasted that in many years of leading teams he had never been successfully sued and did not intend to sully that record with this project. What the … No one around me seemed surprised by this statement (and my head was swivelling around). I couldn’t believe what I had just heard. How many times had customers unsuccessfully sued him? Were lawsuits common outcomes from this consulting firm’s projects? I told my managers at the Alliance immediately and they found it funny too.

A couple of weeks after that meeting the project was cancelled. The Alliance decided to halt the Java project and restart it as a thick client Oracle Forms project. The consulting firm was kept on, just swapping Java devs for Oracle specialists. I was offered the chance to retrain, but decided against it as I believed (and still do) that Java is a more marketable skill than PL/SQL. Thus I was made redundant, but they were very nice about it, and financially it worked out fine. It was an understandable result of the situation - there was no point in having a Java specialist around. Including the month I spent looking for a new job after being given notice, I was at the Alliance only nine months. Although I still keep in contact occasionally with people I met there.