INSTRUCTIONS FOR SETTING UP A MARKETSCAPE EXPERIMENT
December 12, 2000
port: the
computer’s channel that the experiment program will be communicating across
during the experiment
public market: a
market between subjects
private market: a
market between a subject and the experimenter
type:
people categorized by the markets they’re in, their inventories and loans
sub-types:
subcategory of types that have different order sets in their private markets
order set: the
buy/sell orders the experimenter places
GENERAL NOTES:
You cannot set-up production through this interface. To incorporate production you have to make
changes directly to the config file (see Travis for assistance).
Go to http://eeps4.caltech.edu/setup/index.cgi
Try to avoid making mistakes while entering information on
the setup pages, because going back may cause problems or at least require you
to reenter a whole page of information.
If you do make a mistake you can change it afterwards by making
modifications to the config file (see Running
Marketscape Experiments).
On the first webpage:
1)
Choose “Create New Parameter File” to setup an experiment from
scratch, or choose an old parameter file to load settings from an the old file.
2)
Select a source directory.
Usually you will want the “current” one. Any source directory from before 12/9/00 WILL NOT WORK because of
modifications to Marketscape since then.
3)
Set the experiment start date and time - a delay code in the
software will automatically start the experiment at this date and time if you
run the setup immediately (see “setup”
below).
4)
Enter your name in the “Experimenters Name” box.
5)
Pick the “Experiment Type”.
Choosing between “test” or “experiment” sets some defaults later
on. Choose “experiment” only for real
experiments with subjects.
6)
Click the “Select Source and Parameters” button.
On the second webpage:
1)
The First Subject ID should be set to 101 then set the Last
Subject ID to allow the number of participants desired. If you allow many participants this can slow
down the process of loading private markets and cause delays for people
participating in the experiment so you should set this number to the maximum
you want/expect to participate in the experiment.
2)
Observer ID is for anyone who wants to look into the
experiment without participating. This
setting should be left as the default.
3)
Identify the number of “types” to include in the experiment –
you can also choose sub-types of each type later on.
a)
“types” have different inventories of currency and assets,
different loans and different private markets.
b)
“sub-types” have different order sets in their private
markets.
4)
Resources used in the experiment include the currency used and
the list of assets traded, delimited by spaces. By convention this is typed in lower case letters. Only use letters, since other characters
such as the underscore are used internally as delimiters and cause problems. If you want to refresh cash (francs)
inventory between periods, you need to
add the resource “earnings”.
5)
Click the “Submit Part 1” button.
1)
Enter an amount of assets and currency for the
experimenter. The experimenter will
sell or buy units from the participants in their private markets. Remember this amount has to be sufficient to
cover all the orders that people submit from their private markets.
2)
Observer 1 is just an ID for anyone who wants to look into the
experiment without participating. Thus,
the observer is generally not given any assets, loans or currency.
3)
For each type of participant you need to enter the amount of
assets and currency that they start with.
This needs to be determined based on what behavior you are expecting
from the different types.
4)
You also have the option of giving people loans. At the end of the experiment this loan must
be repaid from the money participants have earned. The entry blank for loans identifies to subjects the amount they
have to repay. To actually endow
participants with currency or assets you must still enter something in the
first column (often loans and initial endowments match).
5)
The “reset” boxes give you the option of controlling carryover
between periods.
a)
If ALL of the “reset” items are blank, then all inventory and
orders will carryover periods.
b)
If ANY of the “reset” items is filled in, then all orders will
be cleared at the end of each period. The markets will also be closed for the
time between periods. If any of the
“reset” items has a value, then inventory for that subject for that item will
reset to that value after each period.
You still need to give each type a starting inventory by filling in the
first column of boxes. If you want
participants to start with the same inventory each period you should enter the
same value in the first and last columns. You also need at least one item not
to reset or else the subjects cannot make any money.
c)
If you are planning to reset carryover, you should know the
following details in the configuration file.
The default action is that if
“francs_reset” has a value in the configuration file then the difference
in francs will be transferred to the “earnings” resource after each period. If
you want names other than “francs” and “earnings” then you will need to edit
source files (for now). Either “francs_reset”, or “earnings_reset” MUST be
blank, depending on whichever is the accumulator.
6)
In the “Public Markets” box you should include the names of
all the public markets, delimited by spaces.
By convention, the name of each market is the name of the asset traded
in that market, typed in upper case.
Once again, only use letters, since other characters may cause problems.
7)
Under private markets you can identify the assets which
different types of participants are allowed to trade with the experimenter. These must be associated with one of the
resources that you identified originally.
They should be entered as “Private*” where the * is to be replaced with
the resource traded – by convention, the “P” and the resource should be
capitalized (i.e. PrivateX).
8)
Multiple order sets allow you to change the orders that
participants receive in their private market - this allows market shocks. Different order sets also allow you to vary
the period length (see order sets below).
You can enter as many order sets as you want for the experiment – more
order sets means more data entry.
9)
You can now choose to allow subtypes. As mentioned previously, subtypes have the
same endowment of resources, loans, and private markets but different order
sets in their private markets. Identify
the number of “subtypes” to for each type.
10)
Click “Submit Part 2”.
“Market Resources and
Private Orders” page:
1)
Under “Public Markets”, enter the resource to be traded in
each public market – for simplicity this usually corresponds with the name of
the market, differing only in upper / lower case. As payment enter the resource to used as currency in the public
markets.
2)
Under “Private Markets”, enter the resource to be traded in
each private market – again for simplicity this usually corresponds exactly
with the name of the market. As payment
enter the resource to used as currency in the different private markets.
3)
Under “Order Sets” enter the number of periods and length of
periods in each of each order set.
4)
Under “Order for Private Markets”, you need to enter the
orders that will be placed in the private markets of each type and subtype for
each of the different order sets. This
allows you to set their “value” for different assets. Note that the experimenter
will be placing these orders. “Buy”
means that the experimenter will place a buy order in the private book.
5)
The format of the order should be a space-delimited list of
prices indicating the price that the experimenter will accept for each
unit. For example, if the experimenter
sells the first unit at 35, the second unit at 40, the third at 45, the fourth
at 50, and the fifth at 55, then you should enter “35 40 45 50 55”.
6)
Click “Submit Part 3”.
Final Webpage:
1)
This page allows you to set the parameter save file, the
experiment directory, the experiment URL and the database where the experiment
results are saved. You should verify that the default “Base Port” is not in use
(see below) and choose a different port if it is in use. Everything else should be left as the
default. Remember to record the URL of
the experiment.
2)
Passwords should also be left as the default.
3)
Click “Submit Final Part”
Marketscape setup automatically picks a default port which
is unlikely to be in use. To make sure
that this is actually so, use the “telnet” or “ssh” program to connect to eeps4
(ask Prof. Plott for login and password).
Use ssh if possible, because it encrpyts the connection (unlike telnet),
preventing login and password from being read off the network by malicious
eavesdroppers.
At the prompt, type “netstat -atn”
to get a list of ports in use. These
should appear after the colon in the “local address” column that netstat
outputs. If the default port is not
listed, everything is fine. On the
other hand, if the default port is
listed, you will need to pick another port between 5000 and 30000 that is a
multiple of 100 and not being used.
1) Connect
to eeps4 with ssh or telnet (ask Prof. Plott for the login and password).
2) Change
directory to exp by typing “cd
exp”. You can verify that the
parameter file is correct if you wish.
You can manually correct any errors in the parameter file with a text
editor if you so choose, but it is more difficult (see Running Marketscape Experiments).
3) To
install the experiment type “./inst” and then choose the
number associated with the corrected experiment you just created. Once you install the experiment, you will be
unable to uninstall it.
4)
Once the files for the experiment have been generated go to
the directory associated with your experiment by typing “cd
*” where “*” should just be the date if you stuck
with the defaults.
5)
In this directory start the Perl script by typing “./start.sh”
6)
Once you have started the Perl script the experiment will
automatically begin at the time and date set.
1)
Generic default instructions and announcements will be
appended to the experiment when you setup the experiment. These files are called “instructions.html”
and “news.html” respectively. To change these you can either create a new
webpage or edit the default html files.
a)
For new instructions, create new html files and save them to
the appropriate locations on eeps4 overwriting the default files.
b) To edit existing Instructions or Announcements just use a text editor and type any instructions you want.