I acquired a license to Acqualia's "Soulver" application from a MacZOT bundle a few months ago. The idea is that you can do math in "plain English." As any programmer can predict, this is a total joke.
Q: What is $30 a month for 12 months? A: $360.00 Q: You owe me 50 dollars and 0.62 cents. A: $50.62 Q: M = half a million What is 10% of M? A: 50,000
Wow! Fantastic! As their website says (emphasis theirs): It just knows what you mean.
Let's try some more...
Q: 2 to the power of 2 A: 4 ?: right! Q: 2 squared A: 2 ?: wrong! Q: square root of 2 A: 1.414... ?: right! Q: cube root of 2 A: 1.414... ?: wrong! Q: average of 1, 2, and 3 A: 6 ?: wrong! Q: sum of 1, 2, and 3 A: 6 ?: right! Q: 1, 2, and 3 A: 6 ?: right! Q: quotient of 9 and 3 A: 12 ?: wrong! Q: minimum of 1 or 3 A: 13 ?: mind-bogglingly wrong Q: 1 or 2 or 3 A: 123 ?: wtf
Some of these are extremely fundamental errors. It seems pretty obvious that when Soulver isn't sure what to do, it guesses, and it is willing to be wrong all the time for the sake of making the contrived demos on the web page work.
The idea that some poor sap paid $30 for this is depressing.
I reported these bugs. I have received no response, and there have been no new revs. I guess the definition of snake oil is that if it's selling, it's working.
(I have corrected a mistake or two that I made when transcribing things above, since posting.)Re:eep
rjbs on 2007-07-02T12:12:39
What's really weird is that Soulver uses bc behind the scenes to do its math. I don't really know why. Possibly for high precision, or to take care of variables. It seems like a weird decision, but maybe there's a good reason. This makes me think that this works by applying a series of naive transformations to every line to convert it to bc.
Q: sum of 1, 2, and 3
A: 6
?: wrong!
What would you expect the sum of those numbers to be?
Re:Sum
rjbs on 2007-07-02T12:10:28
Six! Whoops. I clearly zoned out while writing this. Here are some more stupid "plain English" examples:What did happen there? s, 1, x, and s1 are all (no reply). 1x is 1. s1x gets that small number.Q: sum of 5, 6, and 7
A: 18
?: right!
Q: sum of five, six, and seven
A: (none)
?: useless
Q: sum of five, 6, and seven
A: 6
?: wrong
Q: sum of five, s1x, and seven
A: 0.8414709848
?: okay, I typoed; what happened?Re:Sum
ChrisDolan on 2007-07-03T05:01:14
Googling revealed that 0.8414709848 = sin(1)Re:Sum
rjbs on 2007-07-03T12:39:30
Wow... just...That's the right sin(2) for s2x, there. The answer for sin(1) is for 1 degree. The one you gave is for one radian.Q: s1
A: (no answer)
Q: sin 1
A: 1
Q: sin(1)
A: 0.0174524064
Q: s1x
A: 0.8414709848
Q: s2x
A: 0.9092974268
No idea.Re:Sum
jdavidb on 2007-07-03T13:17:00
This was good for a good out-loud laugh this morning. Thank you.
:)
It's kind of funny. If some Perl developer used to the CPAN/Acme conventions wrote such a software, that he admittedly knew that didn't work right but that was supposed to be amusing, it would be appropriate for an Acme module. Instead, they sell it. Wow! that's making money of an amusement (only pleasing for commercial owner's sake and not for user's sake).
