NetNews Usenet Archive 1992 #23/NN_1992_23.iso/spool/sci/math/13289

Text File | 1992-10-15 | 8.4 KB | 341 lines

Xref: sparky sci.math:13289 sci.physics:16620
Path: sparky!uunet!portal!lll-winken!sun-barr!news2me.EBay.Sun.COM!west.West.Sun.COM!smaug.West.Sun.COM!richard
From: [e-mail] (Richard M. Mathews)
Newsgroups: sci.math,sci.physics
Subject: Re: "The Universe of MOTION" (crackpot index)
Message-ID: <[e-mail]>
Date: 15 Oct 92 21:57:22 GMT
References: <[e-mail]> <[e-mail]> <[e-mail]> <[e-mail]>
Organization: Sunsoft Inc., Los Angeles, CA.
Lines: 327
NNTP-Posting-Host: astro
Originator: [e-mail]

[e-mail] (John C. Baez) writes:

>I want the index to be easy to compute.  In fact, what I'd really like
>would be an index that could be computed automatically.

Ask and ye shall receive.  Just before reading your posting, I completed
just such a program.  My simplified rules are:
    100 points each for words in all caps.  To count as a word, it
    must include at least one consonant and one vowel (to avoid the
    words "I" and "A" and to weed out many acronyms).  It must also
    include at least one letter which is not F A or Q.  Digits and
    '@' are treated as lower case letters to weed out login names,
    host names, and more acronyms.

    N consecutive exclamation points count for (N-1) * 100 points.

    100 points for any word which includes a substring which matches
    an entry in the word list which is at the top of the program (see
    below).  Comparisons are done ignoring case.

No negative base value is used on the theory that we must all have some
crackpot in us.

Lines beginning with '>' don't count.  Lines in the article header don't
count except for "Subject:" and "Keywords:".  Subject lines don't count
if the subject begins with "Re:".

The crackpot index is calculated by taking the score determined as above
and dividing by the number of lines examined.  The program prints the
index followed by the name from the "From:" line.

I ran this program on every article in our sci.physics spool (about 2
weeks worth) and then averaged the indices for each author.  The top 10
winners are:

63.0769 Joe Dellinger
64.7059 Khan
65 FRED W. BACH
73.3183 Alexander Abian
87.8875 [e-mail]
107.692 Dani Eder
113.755 John Hagerman
133.333 Scott Brigham, corp
182.353 <[e-mail]>
319.444 Brad Wallet

The prize for best crackpot spoof goes to Brad Wallet.  Hagerman had a
chance at this prize, but blew it by submitting a couple serious articles.
Other sci.physics celebrities (defined loosely as those whose name I
recognized in a quick scan of the list) got the following scores:

2.14634 Paul Budnik
2.56065 Steve Carlip
3.18687 Benjamin Weiner
3.81494 RING, DAVID WAYNE
4.23726 Cameron Randale Bass
5.13109 Douglas G. Danforth
5.66667 Tom Van Flandern
5.79702 Robert Firth
6.61422 Leigh Palmer
6.66667 Dr. Norman J. LaFave <[e-mail]>
7.21700 Ric Peregrino
7.60326 Mcirvin
7.89399 Matt Austern
8.72341 Bronis Vidugiris
8.75751 Daryl McCullough
9.08969 Hartmut Frommert
9.80049 Jim Carr
10.4906 John C. Baez
10.5006 [e-mail]
10.6189 Richard M. Mathews
11.1111 [e-mail]
11.8265 SCOTT I CHASE
15.0000 Matthew P Wiener
15.4365 Blair P. Houghton
18.1966 Mcinnes B T (Dr)
20.1698 Terry Bollinger
25.6695 Jack Sarfatti

There are a few scores which I am disappointed came out so low and a few
which I am disappointed came out so high.  Overall, though, this seems
pretty good.

So here is the program.  It reads a single article on standard input.
The shell script to pass each article through this program and to use
awk or some such tool to do the averaging is left as an exercise for
the reader.

/*
 * crackpot.c   written by Richard M. Mathews   [e-mail]
 * version 0.9
 */
#include <stdio.h>
#include <ctype.h>

char *words[] = {
    "conspiracy",
    "status quo",
    "communication",
    "transmit",
    "ftl",
    "hawking",
    "einstein",
    "newton",
    "galileo",
    "penrose",
    "physicists",    /* as in "all them physicists are out to get me" */
    "scientists",
    "genius",
    "super",    /* superior and super-anything (incl. collider;-) */
    "principle",    /* just seems to be common in crackpot articles */
    "b.s",
    "m.a",
    "ph.d",
    "phd",
    "dr.",
    "provok",    /* variants of provoke and provocation */
    "provoc",
    "arrow",
    "according",
};

char *getline();
extern char *malloc(), *realloc();

char *name;

main(argc, argv)
int argc;
char **argv;
{
    int nwords, i, lines = 0, score = 0;
    char **ptrs;
    int c;
    int all_caps = 1, vowel = 0, consonant = 0, non_faq = 0, bang = 0;
    int debug = 0;

    while ((c = getopt(argc, argv, "d")) != -1) {
        switch (c) {
        case 'd':
            ++debug;
            break;
        default:
            fprintf(stderr, "usage: crackpot [-d] < article\n");
            exit(2);
        }
    }

    nwords = sizeof(words) / sizeof(words[0]) - 1;

    ptrs = (char **) malloc(nwords * sizeof(*ptrs));
    if (ptrs == NULL) {
        fprintf(stderr, "Can't alloc word table\n");
        exit(1);
    }
    for (i = 0; i < nwords; ++i)
        ptrs[i] = words[i];

    while ((c = nextchar()) != EOF) {
        if (c == '\n')
            ++lines;
        /*
         * Check whether this character helps us advance our pointer
         * into any of the words on the word list.
         */
        for (i = 0; i < nwords; ++i) {
            if (*ptrs[i] == c
             || isupper(c) && *ptrs[i] == tolower(c)) {
                ++ptrs[i];
                if (*ptrs[i] == '\0') {
                    ++score;
                    if (debug)
                        printf("%s\n", words[i]);
                    ptrs[i] = words[i];
                }
            }
            else
                ptrs[i] = words[i];
        }
        /*
         * Check whether this word might be all caps (but it
         * must contain at least one vowel and one consonant to
         * count as a word).  Mixtures of caps with digits are
         * not counted as they are likely to be login names,
         * acronyms, etc.  Words connected with an '@' aren't
         * counted as they are probably login or host names.
         * Words made up only from the letters F A and Q are
         * not counted.
         */
        if (isalpha(c) || isdigit(c) || c == '@') {
            if (isupper(c)) {
                switch (c) {
                case 'A':
                case 'E':
                case 'I':
                case 'O':
                case 'U':
                    ++vowel;
                    break;
                default:
                    ++consonant;
                    break;
                }
                switch (c) {
                case 'F':
                case 'A':
                case 'Q':
                    break;
                default:
                    ++non_faq;
                }
            }
            else {    /* lower case or digit or at-sign */
                all_caps = 0;
            }
        }
        else {
            if (all_caps && vowel && consonant && non_faq) {
                ++score;
                if (debug)
                    printf("CAPS\n");
            }
            all_caps = 1;
            vowel = consonant = non_faq = 0;
        }
        /*
         * Check for consecutive exclamation points.
         */
        if (c == '!') {
            if (bang) {
                ++score;
                if (debug)
                    printf("bangs\n");
            }
            bang = 1;
        }
        else
            bang = 0;
    }
    if (all_caps && vowel && consonant && non_faq) {
        ++score;
        if (debug)
            printf("CAPS\n");
    }
    printf("%f %s", lines ? 100. * score / lines : 0., name ? name : "");
    exit(0);
}

nextchar()
{
    static char *buf;
    static body = 0;    /* body of article? (as opposed to header) */

    while (buf == NULL || *buf == '\0') {
        buf = getline();
        if (buf == NULL)
            return EOF;
        else if (*buf == '>')    /* ignore included text */
            buf = NULL;
        else if (*buf == '\n')    /* start of body */
            body = 1;
        else if (body)
            ;
        else if (strncmp("Subject: Re:", buf, 12) == 0)
            buf = NULL;    /* ignore "Subject" on responses */
        else if (strncmp("Subject: ", buf, 9) == 0)
            ;        /* don't ignore "Subject" otherwise */
        else if (strncmp("Keywords: ", buf, 10) == 0)
            ;        /* don't ignore "Keywords" */
        else {        /* all other header lines */
            if (strncmp("From: ", buf, 6) == 0) {
                name = malloc(strlen(buf) - 5);
                strcpy(name, buf + 6);    /* remember name */
            }
            buf = NULL;    /* ignore header lines */
        }
    }

    return *buf++;
}

char *
getline()
{
    static char *buf;
    static int buflen;
    char *bufp;

    if (feof(stdin)) {
        return NULL;
    }

    if (buf == NULL) {
        buf = malloc(BUFSIZ);
        if (buf == NULL) {
            fprintf(stderr, "Can't alloc line buffer\n");
            exit(1);
        }
        buflen = BUFSIZ;
    }

    buf[buflen - 2] = '\0';
    bufp = buf;
    for (;;) {
        if (fgets(bufp, BUFSIZ, stdin) == NULL) {
            if (bufp == buf)
                return NULL;
            else
                break;
        }
        if (buf[buflen - 2] != '\0' && buf[buflen - 2] != '\n') {
            /* need to read more */
            buf = realloc(buf, buflen + BUFSIZ - 1);
            if (buf == NULL) {
                fprintf(stderr, "Can't realloc line buffer\n");
                exit(1);
            }
            bufp = buf + buflen - 1; /* point to the '\0' */
            buflen += BUFSIZ - 1;
            buf[buflen - 2] = '\0';
        }
        else
            break;
    }

    return buf;
}