Reminded of this one by the Rush Limbaugh PDF thread.



  • I was working at a "beltway bandit" on a contract for the USPS in the mid-1990s. The fundamental idea of this contract was to send mail via the internet instead of the postal service, but the postal service would deliver it. The core idea went like this.

    1. Email your document to the post office with a list of addresses.
    2. The post office emails it to the branch offices closest to your recipients.
    3. The branch offices print it out, stuff it in an envelope, charge you some postage, and give it to the mail carrier.

    This is not a hideous idea, but then the implementors got hold of it. Step one was clarified, and "your document" became "your MICROSOFT WORD document"; no other format would be accepted. 

    Step two turned into this monstrosity.

    2a. The post office prints your document at the office which receives it.
    2b. The printed copy is scanned into images. (Not JPEG! TIFF. JPEGs are too small...)
    2c. The images are zipped into an archive.
    2d. The archive is emailed to the USPS branch nearest the recipient.
    2e. The archive is unzipped, and the images are printed.

    There was some pressure from the QA team about this. The argument was that we should simply deliver the original document via email, and that document should be printed at the appropriate postal branch. Unfortunately, not all USPS branches have a copy of Microsoft Word, and significant efforts had already been made to get the scanning process working.

    So instead of buying every branch a $99 copy of Microsoft Word, the plan was to buy every branch a $1400 high-capacity automated scanner. In addition, as will become important later, the team was reduced in size and the test plan reduced in scope. (We clearly had too much time on our hands, if we could critique the design.)

    When this $1400 scanner idea was quite rightly vetoed by the USPS as being too expensive, another solution was proposed: instead of printing and scanning the document, we would print the document to PDF format... but the rest of the process was pretty much unchanged. This was deemed acceptable, and work proceeded.

    After a couple months, we demonstrated a six-page document mailed to a list of twenty-odd people for the USPS reps. The system was designed to process no less than five pieces of mail per second, so the automated envelope-stuffer powered through the hundred-plus pages of mail in less than half a minute. Proudly, the PM grabbed a piece of mail from near the end of the stack and handed it to a USPS rep, who opened it to find only page 4 of the document. Frowning, he looked at the envelope, which bore no address whatsoever... and a postage value of $2.55.

    It seems that our endpoint processing determined how many pages were in each mail by counting the files in the archive, which was now 1 for all cases.

    When presented with a multi-page mail, the system happily printed it out - then put one page in each envelope, and addressed the later pages to other recipients on the list. So the first copy of that six-page mail went to the first six people on the list, one page for each.

    When it hit the end of the list, the system cheerfully continued stuffing pages in envelopes and printing postage on them... but no addresses. (A sane system would stop when it ran out of addresses. This, however, was a government contract. Sanity was thin on the ground.)

    The postage amount, already constrained in the spec to never be over $1.00, was being stored in our system as a single byte. A signed byte. The postage calculation for a NULL address was coming back as -1, which was dutifully interpreted as an unsigned byte because negative postage (like postage over $1.00) never happens. So it printed $2.55 postage on everything that didn't have an address; a little over 100 envelopes. 

    This postage printer was directly interfaced to the USPS network, so it wasn't mere ink on paper - it was actual money the USPS was expected to produce at the end of the reporting period. In less than a minute, we had accounted for over $250 in improperly issued postage.

    In a typically incompetent response, the PM rapidly established that all the postage had been properly debited from the customer's account. The USPS rep asked, after a long and uncomfortable moment, whether the PM actually considered it a positive thing that a customer would have been charged for this inane excess of postage. This was followed by a much longer and much more uncomfortable moment.

    The project was, mercifully, cancelled. The PM pointed his finger at QA for not testing multi-page documents, but certain emails surfaced about which functional areas should be cut from the test pass now that the team was reduced in size...



  • It sounds like a good idea (though horrible execution), but what advantage does it have over sending an email directly to the recipient? The only reason I can think of to send a letter through snail mail is that they won't accept a copy, but this would be a copy anyway.



    Maybe if you were sending to some old fart who still doesn't have a computer, but how long are any of those going to last? The service will be obsolete within years.



  • It's not intended to compete with email, I don't think. Email is no substitute for holding a piece of paper in your hand.

    The whole 'bad data hygiene' method of printing out just to scan back in just gives me cold chills - it's the IT equivalent of eating shit.

    Properly implemented, this sort of thing could cut down on post office fuel and sorting costs in some instances, if you could send a mailing out from one point and the bulk of it was traveled digitally, then you save fuel getting the mail from A to B, leaving the B to C leg a short travel from the printing PO to the recipient. But you would still have to deliver all of the production suppplies to the print offices...and this being a government thing it would never realize the potential savings.



  • Well, the OP said that it was in the mid-90s, when having Internet access at home wasn't anywhere near as common as it is now and plenty of people didn't have email addresses. Maybe they didn't anticipate how rapid take-up would be.

    The largest telco in Australia used to have a service called Telstra easymail where you could dial up a modem to send and receive email - maximum of ten messages at a time, for the cost of a local phone call. No web browsing or anything else. Apparently when it was closed down in 2002 it still had 70,000 customers.



  • @lolwtf said:

    It sounds like a good idea (though horrible execution), but what advantage does it have over sending an email directly to the recipient? The only reason I can think of to send a letter through snail mail is that they won't accept a copy, but this would be a copy anyway.

    Maybe if you were sending to some old fart who still doesn't have a computer, but how long are any of those going to last? The service will be obsolete within years.
    Note the year at the beginning of the OP: mid-1990s.  Back then, email was still rare enough that this idea made sense.



  • I think it does have uses, even today:

    Small Business: Say the need to send the same letter to each one of their customers. It's much easier and cheaper to have the post office automatically print the addresses and put the letter in the envolope than pay someone at their end to do it. Again, for invoices and things that still need to be on paper, it's easier to email a document and address that print it out and put each on in an envolope.

    There is a service out there that acts as a printer, so you don't even have to email it, just File > Print, put in an address and your document's away!




  • @Paddles said:

    The largest telco in Australia used to have a service called Telstra easymail where you could dial up a modem to send and receive email - maximum of ten messages at a time, for the cost of a local phone call. No web browsing or anything else. Apparently when it was closed down in 2002 it still had 70,000 customers.
    And probably a million unread messages.  Thank you spammers!



  • @PeteyF said:

    The whole 'bad data hygiene' method of printing out just to scan back in just gives me cold chills - it's the IT equivalent of eating shit.

    2scanners1paper? 



  • @CDarklock said:


    When this $1400 scanner idea was quite rightly vetoed by the USPS as being too expensive, another solution was proposed: instead of printing and scanning the

    If they balk at paying for $1400 scanners, they'd probably have cancelled the project anyway once they realised that each branch would need a $14000 envelope stuffer and trained staff to operate it for this to work. That's probably part of the reason why businesses which actually do this sort of thing print them out in one place before sending them everywhere.

    And yes, there's plenty of demand for it today. Not so much for the things people want to receive, as for the things people don't yet know that they want to receive. E-mail marketing just doesn't get the same response.



  • @CDarklock said:

    2a. The post office prints your document at the office which receives it.
    2b. The printed copy is scanned into images. (Not JPEG! TIFF. JPEGs are too small...)
    2c. The images are zipped into an archive.
    2d. The archive is emailed to the USPS branch nearest the recipient.
    2e. The archive is unzipped, and the images are printed.
     

    @CDarklock said:

    The postage amount, already constrained in the spec to never be over $1.00, was being stored in our system as a single byte. A signed byte. The postage calculation for a NULL address was coming back as -1, which was dutifully interpreted as an unsigned byte because negative postage (like postage over $1.00) never happens. So it printed $2.55 postage on everything that didn't have an address;

    Sounds like you forgot some steps:

    2f. Overcharge customers.

    2g. ???

    2h. Profit!




  • @lolwtf said:

    what advantage does it have over sending an email directly to the recipient?

    In theory, reliability. Email could fail anywhere at any time.

    Realistically? Postage. Nobody pays postage for email. The post office doesn't like that so much.

    Keep in mind this is in the 1990s, when internet access was still primarily by dialup, and a lot of the country simply didn't have it.

    I did a lot of weird things with the post office. With the spectre of email on the horizon, the post office was scared witless and trying to find new areas of business. I was part of the investigative team - and sometimes a proof-of-concept implementation - for several USPS business ideas. Some of the ones that didn't get out of the gate were still a WTF just for the idea that you would get them from the post office:

    - Telephone service
    - Cable television
    - Video rental
    - Pay per view sports
    - On-demand video

    The most promising idea was catalog ordering. That was a lot like the :CueCat thing many of us will remember receiving in the mail a few years ago. It wasn't much smarter than that, but it was smarter: instead of going to a web site, it actually ordered a product. Instead of plugging it into your computer, you plugged it into your phone, and it sent the data to the phone company who sent it to the post office who sent you a confirmation/bill. Send back the confirmation (via the mail! With a stamp! CHA-CHING), and your order was batched up with everyone else's to get a quantity discount. Catalog retailers were understandably uninterested; print a bunch of barcodes in our catalogs, and we'll sell pretty much the same things we would have sold anyway for a lower price? Pull the other one.



  • @Paddles said:

    The largest telco in Australia used to have a service called Telstra easymail where you could dial up a modem to send and receive email
     

    I remember that. You got an email address like @telstra.easymail.com.au but you had to go to http://easymail.telstra.com.au to retrieve them over the internet (or the other way around). That and the supplied software didn't work properly and was just a waste of time. I also seem to remember being able to buy a device from Dick Smith that would let you view these emails without a computer.

    At that time I had a "TPG Niterider" account which was free dialup between 1am and 7am which suited me fine, as well as the free webmail services were popular. :-)



  • <FONT face="Times New Roman" size=3>What, no wooden table?</FONT>



  • @menta said:

    <FONT size=3 face="Times New Roman">What, no wooden table?</FONT>
     

    The printers were on tables that looked like wood. Except they were actually plastic, with photographs of wood grain under a clearcoat finish.

    Which reminds me, someone told me that the factory where he worked - which made chip and particle board - had been experimenting with how to make similarly strong boards without the particles. Just big sheets of pressed glue. One of the major drives for this was that it would be more recyclable. No easier to use... but easier to throw away.


  • Garbage Person

    @CDarklock said:

    Which reminds me, someone told me that the factory where he worked - which made chip and particle board - had been experimenting with how to make similarly strong boards without the particles. Just big sheets of pressed glue. One of the major drives for this was that it would be more recyclable. No easier to use... but easier to throw away.
    I've encountered this stuff before. It does have an advantage - the same equipment can produce boards of varying strengths depending on the glue you use. The recyclability claim is shit, though - every company that makes the stuff uses a different formula, and thus the "Hey lets just melt the glue and reuse it!" idea falls apart. It's great incinerator fuel, though... Just like all other petroleum-based plastics. (No, really. Incinerator operators are one of the few places that will buy bulk "recycling" plastic because they use it to make the thing burn hotter and thus cleaner)



  • @Weng said:

    The recyclability claim is shit, though - every company that makes the stuff uses a different formula, and thus the "Hey lets just melt the glue and reuse it!" idea falls apart.

    I misspoke. It's not recyclability that was so much better, but compostability.



  • @PeteyF said:

    The whole 'bad data hygiene' method of printing out just to scan back in just gives me cold chills - it's the IT equivalent of eating shit.


    The Criminal Records Bureau, run by Crapita here in the UK, had (5 years ago, but may well still be current) a "quick disclosure" system whereby you rang up, and the operator would take your details over the phone. These would be printed out on a standard form, which was posted to you for signing. Once you sent it back, the form was scanned in by an £X00,000 scanner, which would fail in its OCR approx 50% of the time - at which point, the details would be typed in AGAIN from the form...


Log in to reply