We receive numerous customer complaints involving the Random Play feature of their Compact Disc Player. More owners of multiple disc players complain than owners of single disc players.

We always respond to these complaints and examine the player, but so far we have not encountered any Random function that does not work as designed. Unfortunately, we also don't feel that we have completely convinced our customers that there nothing wrong with their unit. The real problem is that manufacturers are not defining what "Random" means. Therefore, each customer uses their own definition of "Random" and if the player does not behave in that manner, it is assumed to be defective. In our opinion, while it is extremely unlikely that just the Random function could fail, there could be disagreement between your definition and the designer's definition of Random.

Our definition of "Random" Play:

By "scope of play" we mean tracks that are eligible for play. Some units allow random play from a user programmed list of tracks. On multiple disc units this list could span dozens, perhaps hundreds of discs. Some multiple disc units allow the customer to define groups. A "Group" is a list of discs that follow a theme, such as Classical, Jazz, or Rock & Roll. Scope of play can be restricted to discs belonging to a particular group.

Although this definition seems reasonable, there is room for misinterpretation. These differences in interpretation account for most customer misunderstandings.

As a Compact Disc Player design engineer how would you resolve the following issues?

1. Should the player prevent back to back playing from the current disc?

   From a musical variety standpoint this seems reasonable at first glance, but this does change the idea of "random" to mean equal probability of play from any disc except the current disc.

2. Should the player prevent back to back playing of adjacent tracks from the same disc?

   Again, this may seem desirable from a musical variety standpoint, but requires modification of our definition to exclude the next sequential track on the same disc and probably also excludes the prior adjacent track. Should this exclusion be expanded to the next or previous two or three tracks?

   We note that only a person who is very familiar with the specific discs involved will be aware that adjacent tracks have been played.

3. Once a track is played, should this track be excluded until all other tracks within the scope of play have been played?

   While this may also seem desirable for musical variety, there are some hardware implications here. If a player provides this exclusion service, it must have some sort of memory to determine which tracks, possibly out of hundreds within the current scope, have been played. Since memory is costly, and therefore limited, it may be possible for the user to define a scope that cannot be accommodated because memory is full. In that event what will your player do? Turn on a light asking that the user not select so many tracks? Display "ERROR" and stop? Play and mark tracks until memory is exhausted then continue playing random tracks and hope that there will not be too many duplicates?

4. Should the player play a track from all other discs in the current scope of play before playing another track from the current disc?

5. Should the probability of playing a track be based on its length or the length of a previously played track?

6. It makes little sense to a classical music buff to play movements or acts out of order. Would you include an option to allow random selection of discs, but force the cuts of each disk to play in order?

While the above issues are open for debate, there are some practical issues that may override any or all of these points:

1. In a multi-disc situation it is unlikely that all discs in the current scope will have the same number of tracks. If you want to enforce exclusive play and don't wish to allow adjacent track or same disc play, what happens, for example, when the disc with the fewest number of tracks is completely played? Eventually you will be faced with a list of tracks, none of which can be played without violating a rule.

2. In addition to meaning equal probability, "Random" usually implies "can't predict the next selection." For technical reasons we won't discuss here, it is not practical to exhaustively satisfy the "can't predict" requirement. Even big budget research teams use published lists of random numbers or a computer trick called a "pseudo random sequence". There are two important characteristics to note: First, these lists or sequences will eventually repeat. Second, it is possible that the same starting point will be used for the list or sequence of "random" numbers each time the unit is turned on or random play is started. This implies that for a given set of discs and scope of play, the order of cuts may always be the same.

3. In a multi-disc situation the player may not be able to plan the random playing in advance. Planning would require checking each disc within the scope of play. This could take several minutes and require adding more memory. Would your customer be satisfied to wait up to several minutes before the first track is played? Is the customer willing to pay for the extra memory?

As the marketing manager for this Compact Disc Player, how would you respond to the following (conflicting) bits of wisdom:

1. There is a consumer backlash against "complicated" products. Including a large comprehensive manual will cause the retail sales staff and the customer to brand the product as "too complicated" and avoid it. Thick manuals are expensive to create and print.

2. A thin manual, while being cheap to produce and possibly creating the desirable impression of a simple, easy to use product, will actually drain your profits by generating excessive support calls from dealers and consumers seeking clarification.

3. Buttons and features attract attention and sell players. Include as many as you can.

4. Buttons and features scare consumers and confuse the retail sales staff. Include as few as you can.

5. Customers like honesty. If there is a limit to how many tracks can be included in a program or scope, this limit should be clearly explained in the user manual. Including this information in the manual will reduce the number of support calls.

6. Never document a limit, because this will give the competition a point to sell against your product. A magazine reviewer might mention the limit in a review. Besides, the limit is so generous that the consumer is not likely to encounter it. Documenting the limit will needlessly increase the size of the user manual and risk having the player branded as "complicated".

While the above bits of marketing wisdom may seem silly, they must be respected. We can recall a speaker company that built its whole reason for being around the physics of optimum speaker placement in rooms. The owner's manual contained a comprehensive explanation of the possibilities and a range of models was offered to take optimum advantage of each situation. While all speakers and rooms are bound by the same physical realities, we encountered countless consumers who believed that this manufacturer's products were more difficult to work with than other speakers. In fact, these speakers were easier to work with. We are not sure where to place any blame, but it is clear that this manufacturer's message missed it's mark for a significant number of potential consumers. This manufacturer is now out of business.

Our most powerful advice? -- Read your owner's manual.

In addition to reading the section on "Random" or "Shuffle" Play, also read the section on "Repeat". Random and Repeat may interact. Particularly, on multi-disc units there may be an option to restrict Repeat Play to one track or one disc. If the Repeat option is set, Random Play may be restricted to one disc.

All things considered, we feel the rules for Random Play should be very simple and stick strictly to our definition. Most customers would appreciate an option to force exclusive play. (Playing all tracks within the current scope before repeating a track.)

As an example, let's consider the implications of a situation where there are five discs, each containing ten tracks played using our definition of random. We are currently playing a track. What comes next?

• There is almost a 1 in 5 chance of playing a track from the current disc.

• There is a 1 in 49 chance of playing an adjacent track on the current disc.

Although the chance of same disc play is fairly high, we don't feel this is a major problem because only a person who is very familiar with the discs involved and who is paying close attention at the time of play, will notice.

It is possible in an office situation where a multi-disc player is operated day after day with the same load of discs, that the sequence of play will be repeated. A simple solution is to remove some of the discs and reload them in a different order.

It's not likely that any simple or complex mechanized scheme will be able to duplicate your aesthetic judgment anytime soon. If you cannot live with the possibility that a blind random strategy might play tracks in an order that you find unacceptable, manually program the tracks in the order that you prefer.