|
PureBytes Links
Trading Reference Links
|
as seen on TradeLAB list - FYI
rgds hans
------- Forwarded Message Follows -------
Date sent: Sun, 22 Nov 98 16:43:22 -0500
To: TradeLab Mail List <TradeLab@xxxxxxxxxx>
From: Scientific Approaches <sci@xxxxxxxxxx>
Subject: TL_Beta Release
TradeLab is ready for beta testing. Beta testing is very important. It
is analogous to the proof reading of manuscripts by someone other than an
author. Just as proof readers usually spot errors an author overlooked,
beta testers find problems a programmer overlooked. Just as proof readers
often are able to suggest improvements, beta testers often see ways
software can be improved.
There has been only very limited beta testing of certain portions of
TradeLab until now. There are two reasons.
1. Testing has to be repeated whenever code is changed. Large programs
are so complicated that it often is impossible to be sure the simplest
change hasn't adversely impacted something else. Beta testing while major
portions are still being written and revised is largely a waste of time.
2. Communicating with a group of beta testers is a full time job. They
find lots of things they personally would like different. They make lots
of recommendations that are not practical, because of technical
considerations they don't understand, because development costs would
price a product out of the market, because implementation would delay
release by weeks or months , or because implementation would destroy other
capabilities. A lot of time is required to consider, explain and discuss
long lists of such recommendations.
Design is primarily a process of continual compromises. Just as an
automotive designer has to choose tradeoffs between speed, fuel economy,
load-hauling capacity, reliability, comfort, safety, initial cost,
maintenance cost, and many other things, a software designer has to choose
tradeoffs between contradictory software capabilities.
The result in the automotive industry is a range of products optimized to
different applications. Sports cars, luxury cars, trucks, vans, and
motorcycles are not inherently better than one another in a general sense.
They each are best for different applications.
It would be nice to have a single vehicle that would be ideal for Saturday
night, to race at a track, for a family vacation, and to haul lumber for a
weekend project. However, vehicles like that don't exist. We either have
to have several or compromise.
The biggest problem with letting customers help design products is they
each have their own needs, wants, and understandings of particular
applications. If automotive designers released automobiles to the general
public for beta testing, some testers would argue they should be more like
sports cars. Others would argue they should be more like luxury cars.
Others would argue they should be more like trucks, vans or motorcycles.
Good arguments could be made for all those designs, but if a designer
tried to satisfy all those conflicting requirements in a single product,
it probably wouldn't be very suitable for any of those applications.
The same is true with software. I receive private email every day from
people arguing that TradeLab should or must be able to do this or that.
Some of the ideas are good, but many, though not bad if taken alone, would
destroy the overall design for most users. I have spent lots of time each
day explaining the equivalent of why a vehicle can't have the
characteristics of a sports car if it also must have the load hauling
capabilities of a large truck.
Good designers must clearly envision end-objectives throughout a process
of design, at least in a general sense, so the continual stream of
compromise decisions that must be made will optimize toward an end-goal.
Where that is not done, or where portions of designs are significantly
influenced by others with conflicting objectives, the end-result often is
poor. A large part of the problem of dealing with end-user suggestions
during design relate to this issue.
I don't mean to suggest that outside suggestions are not often very
valuable . They definitely are. TradeLab is a better product than it
otherwise would have been, because of some excellent suggestions from this
group. I am only pointing out that there is significant cost involved in
obtaining them, because of the need to make lengthy explanations about why
some are not practical or appropriate. There also is the constant risk of
offending someone who thinks they have a terrific idea that isn't
practical or appropriate to an end objective.
Releasing TradeLab to beta testers while major portions were still being
written would have totally overloaded us with communication. They all
would have had their own ideas about exactly how almost every aspect
should be and every change would have necessitated complete retesting.
Progress would have been nearly impossible.
That would be true even now with a large number of testers. About 350
people want to be beta testers. I get new requests every day. I can't
communicate with hundreds of people generating long lists of bugs,
suggestions and questions each day and make any headway. Even a much
smaller group doing that each day would keep me busy communicating and
leave no time to rework the software to fix problems and incorporate
suggested improvements.
Because of that, we have selected a small group. We hope those who
haven't been asked will not be offended, because no other approach is
reasonable. It would be impossible to make progress with a large group.
Programming knowledge and experience were important considerations,
because programmers more often can determine the true causes of software
problems. They more often can describe software problems with technical
accuracy. They have better understandings of technical tradeoffs and
limitations. Those capabilities reduce the communication workload and
allow more time to improve the software.
We will likely ask a few additional people to become testers as testing
progresses, depending on the problems that are found and the communication
workload with the group we have. Please don't send requests to be
included.
We already have far more volunteers than we can use and I won't have time
to answer.
It is important to keep the beta testing objectives in mind. The primary
objective is to find and correct problems. A secondary objective is to
incorporate suggested improvements according to merit and
appropriativeness.
We know many of you would like to start using the software, but that is
not
a beta testing objective. We have invested a lot in development to this
point. We need to get Version 1.0 finished and on the market as soon as
possible. We can't delay release by months, just so everyone who wants to
can be a beta tester.
I am going to unsubscribe myself from the TradeLab list during the test
interval, so I can devote full time to communicating with testers and
correcting problems they find. Doretta will remain on the list to make
sure it runs smoothly.
It is wonderful to have so much interest in TradeLab. Some of the most
knowledgeable and highly experienced off-floor traders in the world are
members of this list. We are fortunate to have benefited from their ideas
during development (most of which have been sent privately). Some
suggestions haven't been practical or consistent with end-objectives, but
others have been excellent. TradeLab users will benefit generally from
many of them.
That has been wonderful, but my email workload communicating about those
ideas and answering continual questions about the product has been
extreme. It needs to be relieved now, so I can concentrate entirely on the
beta test program until TradeLab is released.
Because of that, I am going to use an autoresponder during the beta test
period that will bounce messages back with an explanation that TradeLab is
in beta test and that no further information will be available until it is
released. Please direct any technical support questions related to
modules that have been released to Doretta at doretta@xxxxxxxxxx during
this period, rather than to me, because I will not see mail sent to me.
We have taken another step to reduce our workload, so TradeLab can be
released more quickly. Our web site descriptions have been confusing.
They describe modules that have become incorporated into the current
product and the current product isn't described. They give estimated
release dates based upon our original plan to release TradeStation add-on
products. There are about a 1,000 visitors a day to the site. Doretta
gets a flood of email every day with questions that take lots of time to
answer.
We continue to receive new orders regularly for the few modules that have
been released. That provides some income, but what we receive isn't worth
the workload from all the questions.
We are working on a new web presentation that describes the current
product and clears the confusion about what is available. However, to cut
the non- stop email overload from the web, we have decided to temporarily
suspend shipment of all modules and simply post a notice that TradeLab is
in beta test, that it is not available yet, and that no further
information will be available until it is released.
You would have to be on this end to truly understand the interest in
TradeLab. It is simply overwhelming. That is great, except we can't
answer questions all day about when it will be released and do the work
necessary to get it released. We made a serious mistake by not taking
down the web site and mail list until TradeLab was ready, right after
deciding to make it into a complete trading program sooner, rather than
later.
The good news is beta testing will start next week. There is no way to
know how long that will take, because there is no way to know what will be
found.
However, I will be devoting full time to whatever is found, so we can get
TradeLab shipping as soon as possible.
I don't plan to post to the list or to read list mail during this period,
but feel free to engage in trading related discussions while I am gone.
-Bob Brickey
Scientific Approaches
sci@xxxxxxxxxx
|