Hacker News Reader

Comment by ErikAugust | Hacker News Reader

You are perpetuating a myth for the sake of a monopoly.

A novel solution could be designed and implemented by a small company but no one dares.

If a small company could design a credible alternative to Google, I'm sure it would have been done - the incentives are too great.

Entrenched businesses are RARELY displaced from their top position. Instead, what usually happens is the world changes around them, and the entrenched business is ill suited to compete in the new world. Nobody ever managed to seriously challenge Microsoft for desktop/laptop OS dominance: https://en.wikipedia.org/wiki/Usage_share_of_operating_syste.... The issue is that desktop/laptop OS dominance is no longer as important as it once was.

i started working on a model, i discovered memex-explorer after and they had apparently designed a slightly less decentralized model but had already implemented it however the project was suspended indefinitely.

memex-explorer + blockchain is the model. A company can not do it. A company can design a platform, and many companies can sell on the optimization & information marketplace to fix problem.

Search, or the quest for knowledge, is not going to be displaced or marginalized. If a small company builds true AI, Google is done.

Depends on how they arrive at it. If they create true AI with lots of spare capacity then maybe Google is threatened, but what seems more likely to me is that true AI is achieved in some hodgepodge system that wasn't well architected to scale. An example of this would be a true AI that came about as a result of embodiment in a robot. Yes, it is true AI but that doesn't mean it knows how to read, let alone read and understand billions of webpages. Even if it could read a page of general text at the level of an average adult, that doesn't mean it can read it quickly.

[EDIT] And as an addendum to this fun tangent, even if the AI was capable, in principle, of reading billions of webpages fast and well, we don't know the power requirements would be for this. This hypothetical small company may have stumbled upon the right algorithms and the right training data to produce a true AGI, but they may simply not have the hardware or the engineering know-how to scale it up across multiple processors. Or scaling it up may require too much (i.e., more than the company can afford) data bandwidth if the robot is controlled. Again, it depends on the details of how they arrived at the AGI.

And, more likely, if a small company does come up with a great AI, they're going to get bought be Google, not compete. Reason being you are still going to need the sheer hardware capacity (a LOT of it) to process so much data. I remember a good article about how YouTube would have likely hit serious trouble had Google not bought them. YouTube was starting to crumble under the exponential growth in load, and Google was perfectly matched to provide infrastructure support.

Maybe. Search is a resource-intensive algorithmic problem. So you need one of two things to beat Google: more resources or a much better algorithm.

You're not going to get the first unless you're Facebook or Amazon or God, but maybe you can build a smarter algorithm. You are up against an army of some of the smartest computer scientists and mathematicians ever assembled -- but what you have going for you is a complete lack of inertia or legacy. You could try crazy things that Google might not, because they won't think it'll work. If you get lucky, one of those blows up. But you have to get very lucky (this is the Innovator's Dilemma in a nutshell).

Anyone want to give it a shot?

You can also niche your space. Hoogle would be an example of that. If you know the searcher cares about Haskell functions only I imagine you can beat Google in that space. That solution probably expands to other interest spheres.

Additionally, you'll still have to find a way to make money. Even if you manage to get on par with Google's search results, it is difficult to replicate their advertising cash cow, and even more difficult to invent a completely new monetization strategy and make it successful.

> ..more resources or a much better algorithm.

Or a limited search space. For myself, it could be HN, SO and Wikipedia.

Content can even be static and downloaded once or regularly.

Use duckduckgo as a first line. Use yahoo and google as backups. Eventually DDG will get enough funding to improve their quality. That's what I do.

I believe the model will be a programatic hierarchical lookup:

local cache ===> personal cloud cache ===> centralized crawl repo ===> small specialized data cache sold into marketplace ===> faiil

There is Yacy (a P2P search engine):

http://yacy.net/

Reply on news.ycombinator.com