[TriLUG] Linux From Scratch

via TriLUG trilug at trilug.org
Tue Mar 4 18:00:31 EST 2025


On 04/03/2025 17:55, William Sutton wrote:
> That said, anyone who has read nontechnical discussion forums should be 
> able to agree that they have a higher than normal noise-to-signal ratio. 
> And I once saw reddit referred to as "the grease trap of the internet."

https://en.wikipedia.org/wiki/Dead_Internet_theory

There are people I know in the (internet) publishing field and they 
seems to agree that there's something to that theory...
Most stuff published on the internet only has one single purpose: drive 
traffic to a page for long enough so that the ad impression counts. 
Whether the content is slop or not, meaningful or not, true or false, 
makes no difference. Their business isn't publishing anymore, it's 
collecting tracking money (what used to be called 'advertising revenue' 
but really is tracking revenue).

> On Tue, 4 Mar 2025, via TriLUG wrote:
> 
>>
>> On 04/03/2025 15:22, Cristóbal Palmer wrote:
>>> On Tue, Mar 4, 2025, at 14:02, via TriLUG wrote:
>>>> Speaking of signal-to-noise ratio, here's a little bash script
>>>
>>> I write for two reasons. The first is to iterate on your script to 
>>> make it 
>> more cross-platform, and the second is to share an example URL that 
>> highlights (for me) the biggest flaw in the approach.
>> >
>>> ....[SNIP]....
>> >
>>> cmp at oof ~ % ./textsig.sh https://www.flickr.com/photos/nasa2explore/
>>> html bytes:   588455
>>> text bytes:     4589
>>> Signal ratio for https://www.flickr.com/photos/nasa2explore/
>>>   .00779838730234257504
>>
>> Oh it was by no means anything scientific, much more tongue-in- 
>> cheek. :) That script is a very, very blunt instrument.
>> It just compares the plain text to the HTML, so stuff like HTML tags 
>> and script and css etc all get considered 'noise'.
>>
>> I was just having some fun right there ;)


More information about the TriLUG mailing list