Hacker News new | past | comments | ask | show | jobs | submit login

> It was dropping for most of the 1990s, and most of the 2010s.

It was dropping in the 90s because of Clinton's enormous budget cuts and firing half a million federal employees which is clearly an outlier and something that now seems to be far more partisan and difficult to achieve. And in 2010s it was dropping relative to enormous stimulus spending and corporate bailouts aimed at helping the economy after the housing crisis.

> So it's been rising steadily over the past 45 years...except for about two decades of that time? This isn't what most people call a steady rise.

You can quote me correctly instead of making up a weak strawman. The entire thread is here for everybody to read, you're not going to be able to fabricate some "gotcha". I said it has been rising steadily for a century, and that recent trend is more shaky but probably still trending upward.




> I said it has been rising steadily for a century, and that recent trend is more shaky but probably still trending upward.

Can you define "recent" here? Because to me it looks like the last 40 years is shaky, and 40 out of 100 is a huge amount.


The comment I replied to which is easy to find is this:

> 1. There are claims that federal spending is out of control. How do you square that with the fact that spending as a percentage of GDP is only slightly elevated compared to the historical average going back to at least the 1970s, with the main deviation in the past few years coming from the after-effects of the pandemic? [1]

Most certainly since 1970 the trend (using a linear regression) has been very significantly upwards, which contradicts that claim.

Since the 1980 the trend has been upwards.

Since 1990 the trend has been upwards.

Since 2000 the trend has been upwards.

Since 2010 the trend has been upwards.

Now just give it a rest for god's sake. Don't keep doing the reddit internet arguing thing.


Separate comment because I feel like it.

If the list of data points was 10 98 times, then a 50, then another 10, it would give you the same results with those linear regressions. You'd see a "trend" going up when you measure from anywhere in the past to the present. But that trend would not represent anything real about the first 98 years. That is not the correct test to use.


That comment was basically right, but I'd say "over 40 years" instead of "1970s" to make the strongest point.

> Since the 1980 the trend has been upwards.

> Since 1990 the trend has been upwards.

> Since 2000 the trend has been upwards.

> Since 2010 the trend has been upwards.

Only one of those statements is true.

Stop using linear regressions for everything.




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: