Skybuck's Payment Method to Develop Software On Time.

[Skybuck Flying]

Hello,

If you have a big software company which doesn't achieve deadlines for
software releases then here is my method to get it done on time the next
time:

Step 1. Give software developers a steady salary.

Step 2. For each second the software is late they pay X ammount of money
back to the company.

Step 3. For each second the software is on time they get X ammount of bonus
money from the company. (Optional).

I am pretty sure this will work wonders.

This will make the software programmer feel involved and will give
incencitives to perform on time if they wanna get rich ! =D

Bye,
Skybuck =D

 

[Xavier Roche]

Le 10/11/2013 12:52, Skybuck Flying a écrit :

I am pretty sure this will work wonders.

Yeah sure, ugly code written as fast as possible is probably the solution!

(Not sure if troll or not.)

 

[glen herrmannsfeldt]

If you have a big software company which doesn't achieve deadlines for
software releases then here is my method to get it done on time the next
time:

A whole book has been written on this problem, called
"Mythical Man Month".

You should read it.

-- glen

 

[Raymond Wiker]

A lot of which has been debunked too.

Which doesn't mean that Skybuck shouldn't study the existing literature
before he comes up with any half-baked[1] theories of his own.

Footnotes:
[1] Possibly a generous estimate.

 

[Robert Miles]

Hello,

If you have a big software company which doesn't achieve deadlines for
software releases then here is my method to get it done on time the next
time:

So you want software that is written well before the deadline but so hard to test that it will still contain many bugs well after the deadline?

 

[bert]

If you have a big software company which doesn't achieve deadlines for
software releases then here is my method to get it done on time the next
time:

You can achieve any deadline at all, simply by not meeting the
specifications on functionality and/or on quality. Indeed, you
can achieve any two of those by not meeting the third of them.
--

 

[Geoff]

You can achieve any deadline at all, simply by not meeting the
specifications on functionality and/or on quality. Indeed, you
can achieve any two of those by not meeting the third of them.

The magic triangle. Cheap, fast, correct. Pick any two.

 

[nick.keighley.zczc]

If you have a big software company which doesn't achieve deadlines for
software releases then here is my method to get it done on time the next
time:

Step 1. Give software developers a steady salary.

Step 2. For each second the software is late they pay X ammount of money
back to the company.

Step 3. For each second the software is on time they get X ammount of bonus
money from the company. (Optional).

I am pretty sure this will work wonders.
This will make the software programmer feel involved and will give
incencitives to perform on time if they wanna get rich ! =D

fix priced contract iff fixed requirements

[I recomend you use some sort of formal specification (I quite like VDM)]
[don't forget to specify timing requirements, memory occupancy etc.]
[better agree the hardware platform in advance]

Note by nailing the requirement down this well you remove some of the benefits of implementing your system in software in the first place

 

[Michael Angelo Ravera]

Le 10/11/2013 12:52, Skybuck Flying a écrit :
Yeah sure, ugly code written as fast as possible is probably the solution!

An old saying "What gets rewarded gets done"

But from "The Official Rules":
"Whatever metric is used to calculate a reward to a person will be manipulated to benefit that person to the point that it no longer serves as a reasonable measurement of the reward due that person."

My personal experience is that you should use "Test cases passed" to set the delivery date. You should assume that the test cases are exponentially distributed in difficulty and required effort and that people will do the easiest ones first.

 

[Mike Stump]

*WAY* Under capacity.

This is like saying you need a faster computer to make the program
that infinite loops run faster. Fundamentally flawed.

Hint, there are those of us that can make a server handle a transation
load of 10,000 a second, and there are those of us that are lucky to
hit 3. They had to sign up 6 people in a week. If their system was
overloaded handling a load of 6 a week, well, let me just say, the
problem wasn't capacity.

The original schedule was likely hopelessly optimistic.

$1B and 3 years? You know, a group of 3 guys, working less than a
week came up with a better web site that does the same thing that
works well, for, well, nothing.

Many people couldn't create accounts.

See, that's the problem right there. We don't want to create an
account, we want to see our buying choices, get information, then just
buy a product if we so choose. An account if fine for something like
iTunes where you want to buy 20 things a month, and not have to
re-enter any details, like shipping address, cc, name.

 

[osmium]

Tech support's
answers for most problems was: "Well, keep trying and it might work
eventually" or "Turn off your pop-up blocker" even after the described
behavior shows that pop-ups are working.

My favorite advice was "Delete your cookies" !!!

 

[Ken Brody]

On 11/21/2013 4:15 AM, Michael Angelo Ravera wrote:
[...]

An old saying "What gets rewarded gets done"

But from "The Official Rules": "Whatever metric is used to calculate a
reward to a person will be manipulated to benefit that person to the
point that it no longer serves as a reasonable measurement of the reward
due that person."

Lines of code. (Comments, blank lines, and #includes don't count -- we're
on to your little tricks!)

#include <stdio.h>

int
main(
int
argc,
char
*
argv
[]
)
{
printf(
"Hello, "
"world."
"\n"
);
return(
EXIT_SUCCESS
);
)

My personal experience is that you should use "Test cases passed" to set
the delivery date. You should assume that the test cases are
exponentially distributed in difficulty and required effort and that
people will do the easiest ones first.

Test case 1: Code compiles without fatal errors.

Test case 2: Code compiles without warnings, using default settings.

Test case 3: Code compiles without warnings, with warnings set to max.

Test case 4: Program links without fatal errors.

Test case 5: Program links without warnings.

....

Test case 42: Program contains zero defects.

 

[Phil Carmody]

On 11/21/2013 4:15 AM, Michael Angelo Ravera wrote:
[...]

An old saying "What gets rewarded gets done"

But from "The Official Rules": "Whatever metric is used to calculate a
reward to a person will be manipulated to benefit that person to the
point that it no longer serves as a reasonable measurement of the reward
due that person."

Lines of code. (Comments, blank lines, and #includes don't count --
we're on to your little tricks!)

#include <stdio.h>

int
main(
int
argc,
char
*
argv
[]
)
{
printf(
"Hello, "
"world."
"\n"
);

Not checking the return value of a function that could fail

return(
EXIT_SUCCESS
);
)


Test case 1: Code compiles without fatal errors.

Test case 2: Code compiles without warnings, using default settings.

Test case 3: Code compiles without warnings, with warnings set to max.

Should be flagged here.

However, your point was made.

 

[Ken Brody]

On 11/21/2013 4:15 AM, Michael Angelo Ravera wrote:
[...]

An old saying "What gets rewarded gets done"

But from "The Official Rules": "Whatever metric is used to calculate a
reward to a person will be manipulated to benefit that person to the
point that it no longer serves as a reasonable measurement of the reward
due that person."

Lines of code. (Comments, blank lines, and #includes don't count --
we're on to your little tricks!) [...]
printf(
"Hello, "
"world."
"\n"
);

Not checking the return value of a function that could fail

Submit a bug report. We'll fix that in the 2.0 release.

[...]

Should be flagged here.

However, your point was made.

 

[jgh]

Step 3. For each second the software is on time they get X amount of bonus
money from the company. (Optional).

If it's on time it's on time. From context you mean "early".

jgh

 

[Mike Stump]

They have signed up a lot more than 6 people in a week.

Ok, the number was a bit more. 5 people per minute, according to
one number (don't know if I should trust it or not) I saw. My point
remains.

The latest figure is something like 28,000 in about 60 days (it might
be an old number reflecting, say, 45 days), and the administration
was saying that's way lower than expected.

Gosh, this this drops us down to 19 an hour. Google handles 59,000 a
second, and that is the result of searching all web sites on the
planet for the exact thing the person was searching for.

Transaction loads aren't the issue when the front page takes a full
minute to load

And there is yet another issue right there. Any decent CDN can scale
to the required they needed. google can engineer a web site that
loads quickly and yet, is usable.

I just timed it, it seemed to take 1/2 a second. So, either you are
wrong, or, or the issue was trivial enough to fix, to have fixed it.

(before you can even get to a transaction). Then they put a "wait"
screen in front

I see no wait screen now. I dived into the site, and didn't see any
waiting. So, seems to be better... The site still gets an Epic Fail
from me, as it is impossible (as far as I can tell) to get any
information (on price) out of the site without playing with email. [
I was wrong, see below. ]

In order to see your choices, you need prices for *you*, right?

No. We need to see the price for the product we select. Might be me,
might be my Mom, might be my daughter, or might be what I perceive is
the standard American family.

*Correct* prices.

The prices displayed by a functioning web site will always be correct,
by definition. Go to www.amazon.com, or www.newegg.com, all the
prices are correct. Like buying a new car, many options and extras
available, give me the base, and then let me refine it til I grow
tired of data entry. People know when they are `done', and know the
price they see before that is a rough guess until they are done.

Well, the prices (including the effect of a subsidy) are determined
by a bunch of information that they were estimating takes half an
hour to enter

It takes 13 seconds. I did it about 20 times in a row on the
California web site. Next. The CA web site is superior to the
federal one, as they give out prices (even though they are wrong if
you have kids). Ah, after you pointed out it was possible to get
prices, I eventually found it (they hid it and made it hard to find,
epic fail). The federal site took about the same 13 seconds to get
the price. The format they show results in sucks compared to the
California one. 1 result per six inches of screen.

and maybe an hour of research on the part of the user to find that
information. Plus they get a whole bunch more info from the IRS.
That part of the design, codified in law, is a major part of the
problem. The law makes computing a correct price *HARD*. What user
would tolerate entering all that information every time they went to
the web site?

We've covered this before. Go to newegg, pick add to wish list.
Notice it just works. Notice it is a single button click. We call
this an existence proof.

As it stands, without logging in, they are giving out price info
(er, lies) for two "ages": under 50 and over 50. That isn't what
you can actually buy, though.

I was unable to get any price [ pause ], oh, wait, I finally found it.
They have improved it, you can enter the number and ages of the people.

Cute, they redirect to http://kff.org/interactive/subsidy-calculator
to actually farm out quote (estimate) calculation, stupid, sad. I
guess $1B isn't enough to figure out the subsidy, maybe $10B would
have got the job done.

When the details you have to re-enter

You predicate your world on the notion that one must re-enter. I
reject that view. A stupid web site certainly can be slow, craps
out, and has you reenter the same information, over and over again. I
view that the same way I would view a car that requires that you get
out and push it.

are more like "type in your last tax return (all of it) and your
credit report", the preference changes quickly. I believe it is also
intended that you come back to the website around Dec. 2014 to get
your insurance for 2015.

I think you misunderstand what open enrollment is, when it starts, and
what they expected people to be able to do. Hint, look it up.

Another indication of a flawed design codified into law: part of the
info required is *estimated* 2014 income. IMHO, asking for estimates
of something in the future on a Federal website ought to be a felony,
as should giving such an estimate.

The IRS does things reminiscent of this. Be curious to learn how it
works out for people that just don't know what their income will be.


Anyway, thanks for your post, it caused me to go try the federal web
site once again.

 

[Philip Lantz]

The IRS does things reminiscent of this. Be curious to learn how it
works out for people that just don't know what their income will be.

The IRS doesn't actually ask for estimates of future income, deductions,
etc., they just require that you pay tax based on those estimates.
Sometimes it really is impossible to avoid paying too much and still be
assured of not having to pay interest on the underpayment. (They call it
a penalty, but it really is just interest, and the rate is not onerous.)

They do have some rules to alleviate the problems caused by inaccurate
estimates, such as: 1) if you pay at least as much tax as you did the
previous year, there's no penalty; 2) if you pay at least 90% of what
eventually turns out to be owed, there's no penalty; 3) if you pay tax
in the early part of the year based on income received so far, and then
your income unexpectedly increases later in the year, there's a way to
adjust for that.

 

[glen herrmannsfeldt]

Mike Stump wrote:
(snip)

The IRS doesn't actually ask for estimates of future income, deductions,
etc., they just require that you pay tax based on those estimates.
Sometimes it really is impossible to avoid paying too much and still be
assured of not having to pay interest on the underpayment. (They call it
a penalty, but it really is just interest, and the rate is not onerous.)

They do have some rules to alleviate the problems caused by inaccurate
estimates, such as: 1) if you pay at least as much tax as you did the
previous year, there's no penalty; 2) if you pay at least 90% of what
eventually turns out to be owed, there's no penalty; 3) if you pay tax
in the early part of the year based on income received so far, and then
your income unexpectedly increases later in the year, there's a way to
adjust for that.

Most of the rules are by quarter. Look at form (and instructions for
form) 2210 to see all the rules.

Most people have withholding, which takes out the appropriate amount
for each pay period. Otherwise, they assume that your income was spread
out equally through the year. If it wasn't, then you file form 2210 to
show that you paid the appropriate amount (estimated tax or withholding)
for each quarter.

-- glen

 

[Michael Angelo Ravera]

On 11/21/2013 4:15 AM, Michael Angelo Ravera wrote:>
[...]
An old saying "What gets rewarded gets done"

But from "The Official Rules": "Whatever metric is used to calculate a
reward to a person will be manipulated to benefit that person to the
point that it no longer serves as a reasonable measurement of the reward
due that person."

Lines of code. (Comments, blank lines, and #includes don't count --
we're on to your little tricks!) [...]
printf(
"Hello, "
"world.">
"\n"
);

Not checking the return value of a function that could fail

Submit a bug report. We'll fix that in the 2.0 release.
[...]

Should be flagged here.
However, your point was made.

My point was that you are probably really about at the half-way point in development when about 62% of the test cases pass.

 

[osmium]

:

<snippage>
This is the most informative thing I've seen on the ACA despite quite a bit
of exposure on various sites and so on. Congratulations!

Whether you have smoked
or not is not an "option".

I hope the actual question is whether you smoke or not. Asking about what
you used to do discourages quitting. The only practice punished by the ACA
seems to be smoking. AIUI a penalty of 50% in premiums. If that is the
*only* thing they punish, they could at least have a more reasonable
penalty, say 20%. It is far from clear to me that a smoker even costs the
system more. A relatively cheap death at 55 vs. a heart transplant later
on, for example. .

Given the subsidy, what's free for one person might cost $1000 per
month for another.

I don't think there is any free coverage. The affordable care act says
_someone_ can afford it, most likely not you. For the short term there is a
subsidy due to the courtesy of the Chinese government.

One must get correct prices to make an intelligent buying decision.
Without that, the site gets an Epic Fail to say nothing of violating
bait-and-switch laws.

I think it is hugely unlikely that the government is subject to the
bait-and-switch laws.