Read carefully a doomsday is aproaching.
If you want a safe y2k stock check out TPRO.
After reading this I got the chills down my spine.
from briefing.com
Stock focus: BMR Software
Daily commentary updated for October 23, 1997
The Y2K software world may soon be turned upside down by a 77 year old man named
Bob Bemer.
Who is this guy?
On the "Software History Center" Website, Bob Bemer shows up in the hall of fame
list as "one of the developers of COBOL and the ASCII naming standard." He also
appears with a notation of "Current whereabouts unknown." Bemer, 77, retired in 1982,
having left his mark in the history books. But if his current company, BMR Software, just
six months old with six employees, lives up to its potential, Bemer's name may go into the
history books all over again, as having the most universal, simple solution to the Y2K
problem, with his Vertex 2000r product.
In an exclusive interview with Briefing.com, Mr. Bemer laid out the workings of his
product and the plans for his company.
What is the solution?
At the risk of being too technical, we present a brief explanation of the Vertex 2000
technology. According to Mr. Bemer, an accurate description of their process has yet to
be made in the media. The following, although condensed, is the first full description
available. If unconcerned with the technical aspect, click here to skip down to an
examination of the business model.
The Vertex solution is based upon the fact that date fields do not make use of the
complete number of 8 bits available to them under IBM ASCII or EBCDIC definitions.
By using the extra bits, new "numbers" can be created that represent centuries. In
combination with the other fields, which are left unchanged, the exact specification of the
decade can be made.
On IBM COBOL systems, characters are contained in 8 bits of data, but single digit
integers never use more than four bits to define a digit. The upper four bits define
what the next four bits mean. But out of the sixteen possible combinations, some were
never used. (Bemer would certainly know; he helped make the definitions years ago.)
Bemer makes use of these unused combinations of bits to identify a number as a
decade-specific BIGIT. For example, in ASCII representation, the top four bits are
0011, (1111 in EBCDIC) meaning that the next four bits are a digit (0-9).
Bemer changes the top four bits of a decade digit (the 9 in a 97 date field) to
represent a decade-century, as a BIGIT number. For example, the top four digits might be
1000 for the century 19xx, and 1001 for the century 20xx. In combination with the next
two digits, a century specific year can be specified in the space of only two characters.
What the Vertex product does is the following:
1.At linkage time, the product scans the code for all arithmetic (addition and
subtraction) operations in the program. Vertex can be included as a library at
compilation with source code, or linked as object code if source code is
unavailable.
2.The product replaces the arithmetic operation with a jump instruction. The replaced
instruction is placed in a table, intact, but the table contains room for additional
information about the operation.
3.The jump instruction takes the program to the table. Execution of the arithmetic
operation is done by the Vertex "virtual machine," not the mainframe hardware. The
Vertex program knows how to add or subtract two BIGIT numbers to get the
proper "year-difference."
4.The resultant values (which are now in BIGIT format) are returned to the program
just as if the original operation had calculated the results. Values returned to a
database are returned as BIGITs.
The beautiful simplicity of the Vertex product lies in the fact that in the IBM COBOL
world, arithmetic operations involve a process called PACKING prior to execution.
Bemer has designed the BIGIT definitions so that packing a single digit character gives the
same post-packed result whether the number is a BIGIT or a digit. Thus, by converting
digits to BIGITS, the packing operation essentially "undoes" the BIGIT information. If a
BIGIT is later acted upon by an operation not under the control of the Vertex "machine," it
doesn't matter; the result is the same. Adding two BIGIT-converted digits with the
mainframe CPU and not the Vertex code gives the same result as adding the original
unconverted digits. "Isn't it simple?" Bemer states.
By storing all of the arithmetic operations in a table, additional information can be
stored there. Vertex has a complex set of logic rules to determine if an operation involves a
date field. When, over time, it is determined that an operation never involves a date field, a
bit is set to indicate this. That operation is then passed directly to the CPU and not handled
by the Vertex code. As the program continues to execute, the Vertex product is able to
build more information into the operations table. The primary purpose of this knowledge
base is to increase performance by avoiding the Vertex code when not needed.
In addition, the Vertex product includes routines for capturing new date fields as
BIGITS when input into a database. This means that input screens do not have to be
changed. In addition, Mr. Bemer is pushing for creation of a new set of fonts which would
allow the printing of single digits to represent specific decades. For example, a 9 (with an
underline under it) might represent the decade 199x. This would allow printing of BIGITs
without alteration of databases, and without altering any output forms.
What is the business model?
BMR intends to use a franchise/license model for distributing its software. By licensing
the code, BMR leverages their product. Potential licensees are waiting at the door for the
product to be finished, according to Bemer. Cost to the end customer is yet to be finalized,
but is currently planned at around $0.15 cents per line of object code. This is significantly
cheaper than the current norm of $1.00 to $2.00 per line of source code, even considering
that source code might condense to object code at a rate of 3 to 1. Converted to source
code, BMR's rate is still only $0.45 per line. This revenue is shared with the license
partner.
Additionally, a significant reduction in testing time is expected as compared to
remediation and recompilation of source code. Since none of the original functionality of
the original program is altered, the only testing required is of results of the operations.
Bemer recommends creation of output data files from original and Vertex-enhanced code
which are then compared. Early testing shows no differences.
Current Status
The Vertex 2000 product has been written. It is currently being tested with two
unnamed clients using their real code. One client, a bank with a $50 million budget, has
hired BMR after unsuccessfully completing a remediation project with another
"well-known" Y2K company, that Bemer politely declines to name. The same code which
was unsatisfactorily remediated will be redone, starting over, with the Vertex software.
Estimated time to install and test is only two weeks. This testing has just begun, but should
it prove successful, will go a long way to validating the BMR concept.
Ship date for Vertex 2000 is mid-December 1997. The company is in negotiations with
several companies to license the software. EDS, while currently unsigned, has examined
the Vertex program and serves as a willing reference for its validity. Transaction
Information Processing of Canada has already signed an agreement to license the software
in Canada. Revenue expectations for 1997, despite the release in December, are
projected at between $5 and $10 million, from clients with contracts who are only waiting
for the completion of product testing. The company is "thinking about going public, but we
want to wait until it is proven that the software works with a client." Mr. Bemer says. This
attitude alone makes BMR stand out from the Y2K tools crowd.
What does this mean for other Y2K companies?
BMR Software is not a public company. Why would Briefing take the time to research
it? Quite simply, because BMR software may set the rest of the Year 2000 world on its
head. Current investors in Y2K stocks should at least be aware of the BMR solution
and the potential risk it poses.
Briefing.com is not in a position to make a technical judgement on the variety of
approaches to solving the Y2K problem. We are, however, in the business of making
judgements on business models and their effects. BMR Software, if the technology
turns out to be valid, has a business model that constitutes a significant threat to current
Y2K stocks.
BMR's business model has three essential elements:
The fee is only 30% of the current going rate.
The costly burden of human intervention is eliminated.
The non-exclusive franchise model means that the sky is the limit deploying the
product.
This model is in sharp contrast to the system integrations or consulting houses who are
working on changing source code. Most Y2K tools are not really automatic and are
mostly assessment based software. The tools find the problem areas, and then humans
verify whether the windowing or date expansion should be applied. (Not unlike a spell
check program.) The human intervention required limits the business potential for many
companies, including tools companies (who mainly license their product to consulting
houses.)
Secondly, the drastic reduction in testing time will appeal to clients. Source code
remediation approaches, in essence, "launder" the code by taking source code out of
production, using a tool to assess, then remediate code (usually involving human
intervention) and then returning it to the user for testing. Testing is required, not so much to
verify the data field changes, but to make sure nothing else was changed. The regression
testing technique involved is time-consuming.
Effect on other Companies
The valuations in the Y2K marketplace are extreme. The promise of fantastic growth in
revenues and earnings has driven the multiples to incredible levels. Some backers of Y2K
stocks go so far as to state that multiples don't matter. But valuations do matter. Any
Y2K stock that falls short, even slightly, of its promise, is likely to see its multiples cut
dramatically, sometimes with drastic effects. Viasoft (VIAS) serves as an example. On
10/21, VIAS reported sharply higher revenues ($13.5 million, up 187% year-over-year
and 19% sequentially) and higher than expected earnings ($0.20 versus First Call estimate
of 0.18). Yet the stock dropped more than 6 points or 13%. Why? Revenue expectations
were for $14.5 million, leading analysts to doubt future growth potential. Y2K stocks
must deliver on the fantastic promise of growth built into their stock prices, or suffer
sharply.
Of course, to do this, the Y2K stocks must win contracts under their current business
model. BMR threatens to change the business model of the entire Y2K industry. Although
it only works on IBM COBOL systems, that is the largest segment of the Y2K market.
Because it is so easy to install and test, and remove if it doesn't work, any large
company is likely to at least try BMR on some portion of their system. If it works,
and the code becomes Y2K compliant in less than a month, BMR is likely to squeeze out
every other vendor.
Bemer admits that the Vertex product won't solve all problems. "We can't fix stupid
programming" he says. "If your leap year calculation was flawed before, it will still be
flawed." But for the far more prevalent case where the problem is simply subtracting the
year 1997 from the year 2001, when all you started with was 97 and 01, the Vertex
solution is ideal.
Of course, if the BMR technology turns out to be flawed, the BMR threat can be
dismissed. It seems unlikely, however, that Mr. Bemer's technology will turn out to be
completely useless. After all, Mr. Bemer, in addition to coining the name COBOL, helping
create the ASCII character set, and inventing the ESCAPE sequence, might legitimately be
considered the first Y2K guru. In 1979, he published an article on the coming Year 2000
problem. At that time, he encouraged everyone to convert to Julian based dates.
"Unfortunately, no one listened back then," he says. When BMR releases its product, they
might listen this time.
|
