Demand-based point-of-use Internet delivery of government documents and forms has the potential for greatly reducing paper waste and government printing costs. The World Wide Web tax forms server is an excellent example of a simple, low cost way to easily accomplish this using an existing telecommunications infrastructure.
Introduction
Taxing Times
1994 was an experiment in the electronic delivery of federal and
state forms and associated instructions via the Internet. These
documents, required for filing of personal and corporate income taxes due
on April 15th in the United States, were provided by using the capabilities
of the World Wide Web. No financial support
was provided by either the federal government or state governments. Disk
space, CPU cycles, network connectivity and power were provided by my employer. All of
my time associated with the effort was pro bono. Many other
individuals contributed to the success of the effort by donating their time
and effort to help me obtain and post the forms. The very first thing I
wish to do is to recognize them, acknowledge their
assistance, and warmly thank all of them as a group.
American citizens will continue to file paper tax returns and other printed paperwork for the foreseeable future. Internal Revenue Service (IRS) projections for the volume of electronic filings versus traditional paper filings show that the vast majority of personal and corporate returns will continue to be paper- based for many years, even with current efforts by the IRS to convert to electronic filing. Additionally, existing legacy systems and resistance to change by the public and professional tax preparation community will require the IRS to support conventional paper filing for many years. Reluctance by the legal community to accept electronic originals and the current inability to eliminate fraud associated with electronic filing also guarantees a long life for paper forms. Of course, the protocols and applications available in the World Wide Web are, once security and privacy objections are overcome, ideal for the implementation of a simple and elegant electronic return filing system.
Access log statistics, presented below, show that a large number of tax forms, instructions and publications were downloaded to numerous Internet sites from Taxing Times over the past ten months. It is, of course, difficult to tell how many of these forms were actually printed and filed with the IRS and state tax agencies. This is the metric against which success of the tax form server concept must ultimately be measured. Some anecdotal information indicates reasonable usage of the forms, however.
For the past tax year, time and lack of any funding precluded doing many of the things which I originally contemplated. Also, a lack of any significant publicity in the popular press for the server (outside of the web community), and the requirement for a (still) fairly high knowledge of computer/Internet protocols and techniques, probably limited it's use to a rarified audience this year.
Based on the large amount of telephonic and electronic mail feedback that I did receive, and what I feel is a successful demonstration of an alternative, cost effective, and resource conserving methodology of disseminating required government paperwork, I am inclined to declare the experiment a success and propose to continue forward with an expanded service for the coming tax year.
Obviously, there were one or more ulterior motives behind Taxing Times. Maxwell Laboratories sells products and services which we wish to advertise in a cost effective manner to qualified potential sales leads. I, and of course many others, hope to maintain the forthcoming "commercial Internet" in a "dignified" form as reminiscent of the original academic and research network. However, we also recognize, acknowledge and support the role that appropriate commercialization of the network can perform in financing and expanding the reach of the future Internet. A public service, provided free or at a minimal recovery cost (as I believe is most appropriate for information and products distributed by the government, and originally paid for by our tax dollars), can provide a powerful "magnet" to draw Internet users to a commercial site. This semi-noble concept was my original impetus in creating Taxing Times. Since inventing Taxing Times, I have been busy creating other public service "hooks" to draw crowds to Maxwell's web site. These include the Missing Children's Database, the Southern California Traffic Report, and the geocoding of http requests. It has also just been a huge amount of fun!
I also must admit that as the year has progressed many times I sit in wonder in front of a terminal window in which the tail of the web server's access log scrolls by in real time; watching the global village come to my virtual doorstep. No previous technology of the information age - including the telephone or television - has provided the potential to interact, even if only in a "virtual sense", one-on-one with so many people so far away.
As the concept of Taxing Times progressed and the potential web site traffic volume became apparent, the profit motive reared it's ugly head and I and management began to plot ways to generate revenue from the concept. This is not an all bad diversion, as the real world has mortgages and nannies to be paid, not to mention support of overhead staff who keep the computers and networks humming. There was a certain drop in the level of "fun" in the effort at this point, however. Various forms of income producing activities were contemplated and some were implemented on a pilot scale. Weekly cost and revenue reports for the project tell a sad tale, however.
This paper covers briefly the documents which were available and delivery mechanisms used for the tax products. Levels of success and problems encountered during the service are summarized, along with some suggestions for future providers of this type of information. Some statistics about total usage of the server and it's most popular forms are displayed. A glimpse of what is planned for the coming year by myself, others, and the government concludes the presentation.
Formats
My first idea was to simply drive down to the local IRS and state tax
offices, grab a copy of each and every form, instruction and publication,
bring them back to work and spend a lot of nights scanning them in as bit-mapped
images for subsequent downloading and use. Although, in
retrospect, given the number of pages, that was a very naive idea, the
results of the scanning exercise were quite good.
IRS Publication 1167 details what the tax agency considers "acceptable reproductions" of original IRS offset printed forms which may be filed. The basic idea is that reproductions which are legible and printed on reasonable (e.g.,bond) paper at close to the original document dimensions are acceptable for filing. Since Xerox (tm) process photocopies are considered acceptable, and since commercial tax preparation software generates laser and ink-jet printed acceptable copies, clearly digital copies at (probably) over 200 dots per inch (dpi) resolution would also be acceptable. There actually was a service the previous year which would fax copies of forms to you (for a fee!) for subsequent filing. Although I don't know, I expect that (knowing what some fax machines are capable of doing to a perfectly legible original) this service was probably not too successful. The IRS prohibits the filing of reproductions of some multi-part forms (such as the W-4 and other employer forms), and frowns on the filing of monochrome copies of the OCR-enabled (optical character recognition) "EZ" series of personal income tax forms. This is mainly because copies of these forms cannot be processed by the existing IRS OCR equipment and thus they must be processed by hand, potentially delaying any refund the taxpayer might be entitled to. Copies of every other form are acceptable, however. (The IRS issued a regulation earlier this year which would have prohibited the filing of copies of the forms for the coming tax year. However, it was quickly withdrawn after the screams of the tax preparation lobby were heard.)
Scanned at 300 dpi as TIFF (Tagged Image Format File) two color images, the IRS black and white forms reprint almost indistinguishably from the originals on a good laser, LED, or ink jet printer. (Most dot matrix printers don't have acceptable resolution.) Using LZW compression in the TIFF format results in image files of approximately 250 Kbytes for each side of a 8 inch by 11 inch form. (IRS forms are usually printed on paper which is slightly smaller than normal bond paper, but reproductions on normal sized paper are acceptable for filing.) Conversion of the TIFF format to a compressed, binary PostScript (tm) format results in a file of comparable size which prints reasonably rapidly on all of Maxwell's diverse collection of PostScript and PostScript-clone printers. Although somewhat large for downloading over an analog dial-up line, the 250 KB files were acceptable for most Internet sites. Forms containing color (such as the -EZ series) were scanned as color photographs at 150 dpi. These resulted in fairly large files, but still acceptable copy quality on a color printer or as greyscale images on a monocrome printer.
Although most of the federal forms were eventually provided in alternative formats to TIFF, I did scan and provide a significant number of IRS forms and instructions in TIFF format. Also, since none of the states were able to provide forms in electronic formats this year, all of the state forms available on the Taxing Times server where scanned and provided in both TIFF and converted binary PostScript format.
Because of the large size of the image files when decompressed (an 8 by 11 inch 1 bit deep image at 300 dpi results in a bitmap of approximately 8 megabytes) some PC image viewers which don't have virtual memory capability have difficulty with these files. However, PC users with limited memory can directly download and print the provided PostScript files with problems.
Not too far into the scanning process, I fortunately became aware of the existence of the entire set of IRS products being available in electronic form. The IRS prints all their products from PostScript format files, and have made these available to the commercial tax preparers for a couple of years. The only drawback to using these is that the forms designers have chosen to use a number of esoteric font faces, which are not part of the normal font collection in PostScript printers, and although available as Adobe downloadable fonts, would cost a considerable amount of money if purchased just to print a form or two for personal filing. Because they are copyrighted font designs, they cannot be included in the PostScript files distributed by the IRS.
Fortunately, this past year the IRS decided to employ the Adobe Acrobat (tm) PDF (tm) (Portable Document Format) to distribute the forms electronically. PDF is a proprietary format created by Adobe to allow the electronic exchange, viewing and printing of fully formatted documents without the need to have every document font available at the receiving end. PDF uses Adobe's proprietary Multiple Master Font technology to recreate very close facsimiles of original font faces using a few generic type styles. (E.g., a serif, sans serif and special symbol font.)
Acrobat files are very compact, as they use a dialect of PostScript which requires a header defining macro operators, such as is used in the Adobe Illustrator (tm) files. This header is prepended to the PDF file when loaded into the Acrobat viewer. This creates a functional PostScript file which can then be saved and/or printed. For all intents and purposes, the resulting printed PDF tax forms appear exactly like the ones printed using the original fonts.
The drawback of this approach is, of course, that it uses a proprietary scheme which cannot be used without a copy of the Acrobat application. Fortunately, the Acrobat Reader (which allows viewing, browsing and printing of the PDF files) was available at a very reasonable cost this past year and it being provided free as of October 1994 by Adobe. Still, use of the application requires either a Windows (tm) PC, a DOS PC with specific graphics hardware, or a high-end Apple Macintosh (tm) and considerable RAM memory and disk space.
Fortunately for Taxing Times users, the Acrobat Exchange (tm) product allows one to "print" the PDF files to disk as normal PostScript files, which can then be sent directly to the printer without the need for the Acrobat application nor the specialized fonts. Unfortunately, because there were over 750 IRS forms, instructions and publications, I was not able to convert all of the forms from PDF and provide them in "raw" PostScript form for use without the Acrobat Reader. I did manage to get most of the popular forms converted by April 15th.
Some of the IRS multi-page publications were only available as SGML (Standard Generalized Markup Language) files. Unable to afford a commercial SGML parser, and unable to find a suitable public domain SGML parser and/or converter, I had to simply place these files on the server as raw SGML. Another problem which was encountered was that the IRS apparently did not provide the associated externally referenced tables and figures along with the SGML text files.
All in all, there are approximately 750 IRS products available on the Taxing Times server. This represents the majority, if not all, of the IRS' current tax year products, and includes some older versions of forms from previous years. Eventually forms were also acquired for about 6 states in time for the April 15th deadline. Most of these arrived through the assistance of individuals who picked up extra copies of their state's forms and mailed them to me. The state of Minnesota is taking a very aggressive tack in trying to provide electronic forms this next year, and they were very helpful in providing me with a complete set of forms to scan in for this year. I managed to get all of their forms on-line, but did not have time to finish the complete set of instructions. Since they were local, I also managed to get the entire set of California state tax forms and most of the instructions on line in TIFF and TIFF-converted PostScript formats.
Delivery
I originally conceived of three delivery modes of the tax products
available in Taxing Times. Clearly, the World Wide Web capability of
downloading to disk a clickable link is the first. This seemed to have
worked quite well for most users. I tested downloading all three tax form
formats on Unix, Macintosh and Windows PC platforms using the then-available
versions of the NCSA Mosaic
browsers. All worked locally, with the exception that the
Windows version of the browser was quite immature at the time. Others
reported success in downloading with the Lynx and
Cello browsers. Downloading the TIFF images directly into an
external graphics viewer (JPEGView or
GIFConverter on the Macintosh, and Lview on the Windows PC) worked well
under Unix, sometimes on the Macintosh, and I was never able to find a
viewer on the PC capable of handling the 8 meg files. Others must have
had better luck, as many TIFF images were downloaded. Downloading the
"raw" PostScript files to local disk and then printing them to a
PostScript printer seemed to work almost without fail. Some users
reported problems with printing the files on some PostScript printers. I
found that PostScript files which had been converted from the PDF format refused
to print on our RealTech printers (which actually have real Adobe
PostScript Level 2 interpreters), but printed fine on even old Level 1
PostScript clones. The final format available on the Web server were the
PDF files. These downloaded best, being the smallest, but required the
availability of the Acrobat application on the local machine in order to
view and print the files. As Acrobat was not available until this past
summer for a couple of Unix workstations, this was a problem for site
users who had only Unix machines available. Since the NCSA Mosaic
browser can launch an external application based on file name extension
(e.g., ".PDF"), users could configure their web browser to automatically
launch Acrobat and display a downloaded PDF file. This worked quite well,
provided enough RAM memory was available to have both Mosaic and
Acrobat resident at the same time.
The second delivery mechanism which would have been useful was an anonymous FTP site containing the tax products. Unfortunately, security paranoia and a lack of time and funds precluded offering this service this year. The third delivery mechanism, and one which I had high hopes for, was the automatic e-mailing of tax forms back to a requestor. I installed and configured Johan Vroman's (jv@mh.nl) mail robot, the Squirrel Mail Server to automatically mail out a tax product based on a user's request for a specific document number. I limited this service to sending out of only the PDF files in order to minimize the number and size of the return mail messages. The mail robot automatically parses the user's request, finds the necessary tax form file, compresses and uuencodes it, and then breaks the file into the necessary number of separate e-mail messages to fit within the limitations of the Internet mail system.
Experience and Statistics
Taxing Times went on-line in various stages beginning around January
11th, 1994. The bulk of the forms library was in place by mid-March.
S-Cubed switched over from a 56 Kbyte/sec Internet connection to a T1 link
around this time, fortunately just in time to accept the flood of accesses
in April. Our http server runs on an SGI Indigo R-3000
workstation, which shares a complex of Ethernet segments with
approximately 100 other workstations and a few hundred PCs and
Macintoshes. The bandwidth of the 56 Kb line saturated at around 300
access per hour. This, I believe, was evidenced by the large number of
dropped connections seen in the server's error log file. Once we
transitioned to the T1, we peaked out at around 1500 accesses per hour
the week of April 15th. The limit here was apparently the I/O subsystem
of the Indigo, as our logs of the T1 line show that this amount of traffic
was using less than 5% of the 1.44 Mbyte/sec bandwidth on average. The
traffic on the morning of April 15th required rebooting the Indigo
numerous times and effectively precluded anyone else from doing any
work on our LAN that day. This was caused apparently by the loading on
the Indigo (which provides boot service for the other Indigos) and not by
the network traffic, however.
A total of approximately 40,000 tax forms and documents is PDF, PostScript and TIFF formats were delivered between January and September 1994. The statistics show that the raw PostScript format of the forms was the most popular, and although significant benefits are seen in the PDF/Acrobat delivery format, for simplicity and ease of use plain PS is clearly the best and most accepted. TIFF should probably be used only as a fallback for forms which are not available as original electronic documents, and even then should be converted to PostScript for downloading and use. The breakdown between the three formats is shown in Figure 1. The http server logged over a quarter million http requests during that time. Over 10,000 unique Internet addresses visited the site. I received over 1500 e-mail requests for downloading instructions and over 400 requests for tax forms via e-mail. One log file for the three weeks following April 15th is unfortunately missing, so I have estimated the traffic for this period based on the next available log file (assuming that traffic was minimal after the 15th).
Since most Internet soothsayers contend that the majority of users still don't have WWW browser access the network, I also wanted to provide at least one mechanism for delivery of the forms to users without the need to use the WWW. Clearly, one would like to provide access to the forms using, with the exception of telnet, all the standard Internet transfer protocols (http, gopher, ftp, and mail).
Time, monetary limits and security paranoia precluded implementation of gopher and ftp servers this year. I attempted provide the mail robot to deliver forms upon request automatically, but was stopped by security worries at the time over dangers in the Unix mail program. I tried to answer, by running the robot manually, as many of the forms requests as possible. Given the volume of requests, I rapidly fell impossibly behind and failed, however. Practically, too, because most of the forms were necessarily split into multiple mail messages, it was difficult for a novice Internet user to re-assemble, decode and decompress the forms for use. Subsequent adoption of the MIME mail attachment standard and the availability of robust Internet e-mail software (such as Qualcomm's Eudora) make this concept much more viable now.
As mentioned in the Introduction, it is difficult to gauge the actual usefulness of the effort this past year without somehow tracing the number of Internet-delivered forms which were actually filed. Obviously, the "inquisitiveness factor" of the part of Internauts arguably resulted in the downloading of hundreds, if not thousands of forms "just to see if it works" which were consequently never used. I believe that some general statements can however be made about the usefulness of the server to the public (at least to the Internet public at this point). Some of these statements can actually be substantiated by feedback received from Taxing Times users.
Next Year
The FedWorld and the IRS are currently
finalizing an agreement to allow FedWorld to supply the IRS products next
year. Consequently, I reluctantly, but with some relief, hand "my baby"
off to them and wish them the best of luck. An official announcement will
be made in December, but I can unofficially relate that FedWorld will
carry all of the IRS products in Acrobat PDF format, PostScript, HPGL (tm) and
SGML formats. Some instructions and publications will also be available
as ASCII text. The Adobe Acrobat 2.0 Reader will be available for free
downloading. Although I forgot to ask, I assume that FedWorld will
provide the tax products free of charge -- the do however charge for some
of the government products they carry. Forms and other products should
start becoming available on the server in December. FedWorld offers
gopher, WWW, ftp and dial-up BBS access.
Since there will be significant publicity for the FedWorld service next year I expect a large amount of traffic to the FedWorld server in April 1995. I have suggested to them that they consider spawning mirror sites as April 15th approaches in order to make sure that Internet users are able to get the forms they request.
For next year, Taxing Times will continue to provide it's services to the Internet community. If feasible, we will mirror the IRS products from FedWorld. In addition, I hope to provide a central meta-site with links to existing state tax form servers as they (hopefully) come on-line. I will, if necessary and possible, provide federal and state forms again. There is significant interest in providing appropriate Canadian forms and instructions also in 1995.
Our major project for the coming year will be to provide the IRS and state instructions and publications in hypertexted HTML format. We hope to provide the entire tax code - 26USC - also in this format. I believe that this will be most beneficial, as users will be able to search for and browse pertinent sections of the instructions without having to download and print them.
Obviously, the day will soon arrive when the capabilities of the web browsers and servers can be used to actually complete and file one's tax return completely on-line. The certain availability of secure browsers and servers, along with the availability of digital cash, will make this a tremendous growth area for the professional tax preparation community. There will be no need to buy a tax preparation program, as you'll simply connect to (for instance) Chipsoft's web server, provide a credit card number to charge the service to (or some digital cash), and begin filling out the forms using their CPU to do the calculations. Filing will be electronic and Chipsoft will float you a loan against your refund in the form of digital cash, which you'll then be able to use on the Las Vegas WWW server!
Acknowledgements
I would like to acknowledge the assistance of numerous individuals,
without whom Taxing Times could not have happened this past year. Dr.
Kerry Dance for his vision and support of the concept. Mr Dan Solari of the
IRS, who represents the type of federal employee who destroys popular
stereotypes. Ms. Cathy Bauman and Ms. Debra Kudeviz of Adobe Systems.
Matthew Pulliam, Stephen Hollenhorst, Somesh Rao, and Dave Carter, who contributed
state tax forms. Dave Proffer, my brother, who supplies
equipment when my company can't or is unwilling. And Jill Vorhaus and
Dale Dougherty of GNN/O'Reilly
Publishing, for recognizing Taxing Times this year and thus greatly
raising the visibility of the concept.
About the Author
William Proffer is a Senior Staff Scientist in the Applied Technology Program at
the S-Cubed division of Maxwell Laboratories. Maxwell is a high technology
research and development public corporation headquartered in San Diego,
California. He has been with S-Cubed for the past 17 years working
in computer science and geophysics related areas.
Mr. Proffer is currently involved with various High Performance Computing projects using local computing resources and remote resources available on the DREN and NREN. His World Wide Web projects are supported by company computing resources, but have been done primarily on his own time.
He is a member of the International Connection Machine User's Group Executive Committee. He was a participant in the Winter 1993 DoE/LANL Invitational Computational Science Workshop at the LANL Advanced Computing Laboratory and was a visiting researcher during 1985 in the Computer Science Department at Carnegie-Mellon University.
Mr. Proffer received his Master of Science in Applied Mechanics from the University of California at San Diego, Department of Applied Mechanics and Engineering Sciences and his Bachelor of Arts in Applied Mechanics and Engineering Science from the University of California at San Diego, Revelle College. He resides with his family in Escondido, California, USA.
