HackerTrans
TopNewTrendsCommentsPastAskShowJobs

cosheaf

no profile record

Submissions

Mathematics of Collective Intelligence

ipam.ucla.edu
2 points·by cosheaf·4 lata temu·0 comments

Second years are my most active [drug] customers

theboar.org
3 points·by cosheaf·4 lata temu·0 comments

Climate Change: Annual greenhouse gas index

climate.gov
1 points·by cosheaf·4 lata temu·0 comments

Caught Up in Complexity (2021)

dothemath.ucsd.edu
2 points·by cosheaf·4 lata temu·0 comments

The Parable of the Paperclip Maximizer

hackernoon.com
1 points·by cosheaf·4 lata temu·0 comments

If We Don't Fix Sensemaking, We Won't Survive

youtube.com
2 points·by cosheaf·4 lata temu·0 comments

Presheaves as co-limits of representable functors [pdf]

github.com
2 points·by cosheaf·4 lata temu·0 comments

Sets for Mathematics

cambridge.org
2 points·by cosheaf·4 lata temu·0 comments

On Folk Theorems (1980)

dl.acm.org
8 points·by cosheaf·4 lata temu·2 comments

Silicon Valley billionaires believe in the free market, as long as they benefit

theguardian.com
4 points·by cosheaf·4 lata temu·4 comments

Our hunter-gatherer future: Climate change, agriculture and uncivilization (2020)

sciencedirect.com
9 points·by cosheaf·4 lata temu·0 comments

[untitled]

1 points·by cosheaf·4 lata temu·0 comments

A Cross-Modal Audio Search Engine Based on Joint Audio-Text Embeddings (2018)

microsoft.com
2 points·by cosheaf·4 lata temu·0 comments

PyTorch Lightening

pytorchlightning.ai
1 points·by cosheaf·4 lata temu·0 comments

Spectral Audio Signal Processing

dsprelated.com
2 points·by cosheaf·4 lata temu·0 comments

[untitled]

1 points·by cosheaf·4 lata temu·0 comments

[untitled]

1 points·by cosheaf·4 lata temu·0 comments

comments

cosheaf
·4 lata temu·discuss
Maybe I'll make a support page and then you'll read it and believe it.
cosheaf
·4 lata temu·discuss
There is no royal road to changing one's mental behaviors and habits. It's a metacognitive skill and can be developed like any other skill but you'll have to put in the work to develop it. Start by developing an exercise routine that you can stick to, preferably cardio. Then make sure your diet is not full of junk calories. If you manage those two then you'll be on your way to developing higher level cognitive skills and might resolve whatever issues you have with your OCD.
cosheaf
·4 lata temu·discuss
Correct and if you deviate from the standards then you will be medicated until you are subdued and willing to work for the machines destroying the habitability of the planet.
cosheaf
·4 lata temu·discuss
Maybe you should have lined up projects before buying them.
cosheaf
·4 lata temu·discuss
Chocolate is better and cheaper than coffee, and might even be better for overall health in the long run than coffee. Whenever I'm at Trader Joe's I make sure to buy whatever is on sale and is as close to 100% dark chocolate as possible.
cosheaf
·4 lata temu·discuss
You should sit down and think about how Google makes money instead of reading whatever is on their support page.
cosheaf
·4 lata temu·discuss
Because that's the correct intuition. The contemporary world is not designed for sane people. All the systems around us are designed to alienate people from each other with artificial intermediators like money, status, fame, and just generally useless toys and gadgets. To succeed in such a world you have to have sociopathic tendencies. You have to value material goods and possessions over people which is kinda insane when you really think about it.

There is an interesting interview with Brit Marling addressing some of these points about her time working at a bank (Goldman Sachs): https://duckduckgo.com/?q=brit+marling+goldman+sachs&t=ffab&....
cosheaf
·4 lata temu·discuss
That's reasonable. Substitute "corruption" wherever I used "gaming" when referring to maximizing profits at the expense of serving useful search results.
cosheaf
·4 lata temu·discuss
This is actually what google does. They find the 50 sites for your query and then do a multi-armed bandit test to see which one gets the most clicks but with a bias towards sites that serve ads from the Google ad network. A search engine without that bias is not gameable because it will converge on the results that is most popular for a given query and not because it also serves ads that affect the search engine's bottom line.

Popularity is gameable but not the same way as Google is currently gameable because as soon as a site becomes popular and starts exploiting its ranking it will be easy enough to add a decay factor to prevent such sites from dominating the top results during the multi-armed bandit stage of ranking.

In any case, the logic of why Google is going to shit is obvious. Arguing about fixes is not going to change their underlying business model and why spam is dominating their results. As long as they are a search engine, an ad network, and a corporation that must maximize profits their results will continue to deteriorate until the top results are all just spam.

> I don't think it was a pedantic argument or a misunderstanding. I think Lascaille understood your position and disagreed with you, not just pedantically but over the substance.

Then that wasn't clear and seemed like a pedantic point since it's obvious that any algorithm is gameable and I should have made it clear that I wasn't talking about a perfect search engine but one that was not susceptible to profit driven spam (which is currently the reason that Google results are going to shit).
cosheaf
·4 lata temu·discuss
Thanks but it's not really my analysis. I learned it from an art project: https://googlewilleatitself.com/.
cosheaf
·4 lata temu·discuss
I'm not interested in a pedantic argument. If you didn't understand what I meant then you should have asked for clarification. A search engine designed to surface useful information is not gameable if it is not in the business of generating quarterly profits from its own ad network. A site designed to drive traffic to itself can still try to hack the system by generating spam but without Google's incentives for surfacing such content because it serves ads from its own network there will be fewer such sites and useless content to go along with it.

At the moment Google is incentivized to uprank spam because the spam comes with ads from its own ad network.
cosheaf
·4 lata temu·discuss
We're in agreement. I don't think quarterly earnings are the right way to design and build products. Maximizing profits is not correlated with value and is often inversely proportional to it. Google was so successful that they changed the incentive structure of all content on the web. Now they're like the yeast drowning in the byproducts of their own metabolic processes. They exploited whatever nutrients were available (hand curated links) to make them initially successful and now there is no more worthwhile content being generated that is not designed to rank highly on Google (which is not the same thing as quality content and is instead content optimized for generating ad revenue from the Google ad network).

It's the same with social networks, upvotes and likes skew the the type of content that is generated to be liked by the people liking and upvoting instead of being insightful. Popularity is not the same thing as insight so the social web is full of mostly useless but popular content.
cosheaf
·4 lata temu·discuss
Yes, it's very similar with the added caveat that Google has an interest in serving results that have ads from their own network. This is why Google's metrics can be hacked. Anything that is barely above being classified as spam but serves ads from Google's ad network will be prioritized over other results simply because they have to hit their quarterly revenue targets. SEO hacking is not possible if a search engine is just a search engine but Google is also an ad network so they will always be susceptible to being gamed.

This is also the case for social media platforms. They're incentivized to surface content that generates engagement and ad revenue. Basically ads are at the root of all problems when it comes to the internet and the content on it.
cosheaf
·4 lata temu·discuss
Which is why I said it's an ouroboros eating its own tail. The scale it operates at and given that it's also an ad network guarantees that whatever results it finds will favor its own ad network. It doesn't even have to be intentional since all they're doing is optimizing some metrics and running ML algorithms. There is no single person that could be blamed for the deterioration of the results. There is no way around this conflict of interest and they will continue pushing the envelope to increase their own revenue at the expense of useful results for as long as possible.

If a site is hosting ads from the Google ad network and is barely above being spam then Google will prioritize it over other results in order to maintain its quarterly revenue predictions.
cosheaf
·4 lata temu·discuss
It's essentially a machine for printing money and people don't really understand what they're giving up in exchange for "free" search results. Google is beholden to market forces, it's no longer in the business of indexing useful information because the market doesn't value useful information, it values ad revenue.

This is a structural problem and anything that gets large enough will succumb to the same forces. If the incentives are for optimizing ad revenue then that's what all corporate machines will do at scale, regardless of their initial motives and incentive structure. It doesn't help that Google is also an ad network, hence the ouroboros aspect.
cosheaf
·4 lata temu·discuss
So why do you suppose Google doesn't surface the domains and results you presented? For example, I searched for "game reviews" and gameboomers was nowhere to be found.
cosheaf
·4 lata temu·discuss
Great, now compare this to all the content generated on content farms for selling ads and gaming search engine rankings.
cosheaf
·4 lata temu·discuss
Show me a site made by people with manually instead of algorithmically curated links and content.
cosheaf
·4 lata temu·discuss
Federation is the only reasonable solution at this time but the technical overhead of federated search is high enough that most people won't use it so it won't benefit from network effects like Google did in the beginning. There might be a combination of blockchain juju that could make federation viable but all the thought leaders in that ecosystem are too high on their own supply to realize they could use blockchains for anything other than gambling.
cosheaf
·4 lata temu·discuss
They did that right when they chose the advertising model. It was never going to work in the long run and the founders knew it. They just thought they could build an AI system before that happened and it turned out they were wrong. Useful AI that could distinguish real knowledge from SEO optimized spam was much further away than they thought/imagined.