General discussion

  • Creator
  • #2189150

    Dumb is good!


    by leoofmars ·

    Dumb numbers are good. You can?t run out of them and computers can assign them.

    A ?dumb number? is a number which does not contain any meaningful internal elements. It?s just a number. A smart number, like the Vehicle Identification Number on your car, tells you all kinds of things such as the manufacturer and who tightened the bolts on the brake pad.

    When in comes to identifying equipment in our new ERP, our Maintenance Department keeps insisting on using a smart number that will tell you the location, the process involved and several other pieces of information, because ?people need to know this when they are in the field?. They seem to be oblivious to the arguments that:

    Anyone looking at a piece of equipment in the field will immediately be able to grok all this info by turning their head from side to side and processing visual data, and
    Anyone using the ERP system will be able to look at related fields and not need to know the secret codes embedded in the number.

    During an hour and a half discussion of this yesterday ? which started going around in circles after the first twenty minutes ? someone finally asked, ?Where do they (maintenance) get these numbers they want to use? They don?t relate to anything else we?re doing now.? Unfortunately, I?m the only one old enough to know the answers. The numbers come not from our current system, but from a custom program written on an IBM S/36 which the Agency stopped using in 1993.

    Has anyone run into this smart number fetish before? How do we convince these gentlemen, who are not dumb, that all they need is a dumb number?

All Comments

  • Author
    • #3071812

      Right ON!

      by dr dij ·

      In reply to Dumb is good!

      I setup PC numbering scheme, and left off location, etc. as PCs move then you have to delete the # and put it in under new#. Instead ‘dumb number’ or unique ID relates to nothing except fields about the PC which can change when you move it. with their ‘smart #s’ (01WA09) for pc #s, for example, our PC dept can move the pc out of ‘WA’ dept and re-assign it then have big hassle as what to do, re-assign # and screw up it’s history? People need to separate info about an object from its identifier.

      • #3071742

        it depends…

        by john.a.wills ·

        In reply to Right ON!

        … on what kind of information is in the smart number. I recently constructed a database for a locksmith, and the main table was not of keys but of keyings, i.e. of actions to attach a key to a door. Dumb identifiers would have allowed duplication, but the smart identifier included the date, which cannot change for a particular keying, and 3 other items determined by the keying action.

    • #3071641

      You keep the dumb number

      by tony hopkinson ·

      In reply to Dumb is good!

      to yourself and expose the smart number, by building it from it’s parts for display/reporting.

      Smart numbers occur a lot in asset numbering schemes particularly ones from a historical paper based system.
      Real example
      0035/014/0200/356 might not mean anything to you but I know a guy who could access the part it refers to from the directory structure on the CAD system, and find it a hardcopy in a set of very large cabinets full of A0 paper.
      It’s not can you get rid of it but why should you. Please don’t tell me it’s an so you can use an identity column in a database.

    • #3088532

      Codification and the Human Mind

      by thomgordon ·

      In reply to Dumb is good!

      The use of dumb numbers, IDs or assession numbers is the thing of legends for DBAs. Unfortunately mere mortals feel more comfortable with numbers they can relate to.

      So how do we strike a balance between ‘smart’ technology (using dumb numbers) and the need for humans to identify things from the numbers or codes assigned?

      Simple – we build both. As programmers and database experts all know, building codes into numbers leads to death whenever we hit a limit. So why not assemble a number for your users using the various codes that define the attribute the dumb number represents.

      Oracle applications, for example, uses tragically stupid numbers under the covers but can blend together useful codes to form the numbers that real humans read and use.

      The VIN, for example, is composed of various attributes assigned to an automobile. Some manufacturers use slightly different codification but it all comes down to a code that can describe the automobile to someone who knows how to read it.

      The killer with the VIN is that you have to cheat when you BUST a limit (single digit year!). The secret codes can help with many real-world task. The trick is to remove the secret. The Publishing Industry just stretched their Secret ISBN to 13 digits from 10 to help adapt to more books. Not a problem if the ISBN is just a code and you identify the books by a ‘dumb number’ under the covers. True you still need to stretch all your applications but you only need to do conversion of one number in the database. If you entered the wrong ISBN number by mistake, you don’t need to re-create hundreds or thousands of records throughout your database to correct the problem.

      You solution though, is to embrace the needs of your users and not discount them. Find a way to help them identify their ‘real-world’ STUFF so they can get the meaning they need. It truly is handy to know you are looking for a 1/4 inch nut and not a Lambourghini…

      Good Luck

Viewing 2 reply threads