An algorithm could be developed that could track the trends of human spam-sniffers (taggers?) e.g. key words could be harvested from spam marked by human sniffers and then could be used to automatically sniff out and block trending spam. This algorithm could then pre-select and block the spam as it trends automatically. It could also be open source code, so everyone knows that it’s not been modified to allow for certain spam.
This algorithm could thusly provide a cheap counter to cheap spam. Elite spam would still get through, but the paid human sniffers would get that.
I don’t think there’s much disagreement between us about the viability of spam-free search engine.
I don’t say it won’t ever work, I just said I expect that initially (say up to 2 years following the launch) it’s going to be a big problem, and later as things fall in place (e.g. reputation systems, (possibly voluntary and verifiable) user ID’s, etc.).
And my expectation is not MaidSafe-specific, it applies to other similar crypto technologies (not necessarily related to data sharing). I am aware of new Bitcoin-based approaches to IDs and reputation, but those are just being created, and even for the Bitcoin network it will be at probably a year or longer before any of them gets used by more than 5-10% of users. We should also consider that many users have no interest in using those to begin with, so to ensure voluntary participation of “taggers”, it’ll have to exist, provide sufficient privacy and functionality, and work well with search engines.
That is at least 16-24 months away.
I was thinking the same. I’m not sure if @physics was actually thinking on build this or was just giving an idea. I hope he is taking the initiative
But I think he will not get upset if you start your own search engine on safe.
My 2 cents in case you start something like that:
you can start a kickstart project to fund it;
this kind of project will require a front/back end design - I can’t think on this working as a serverless app;
client will send a message with his query and he will receive a message with the results - just like google, but with safe messaging instead of http/tcp;
backend can be decentralized (maybe on a “phase 2”): create your own alt-coin (searchcoin) + allow anyone to download the mining app + the mining process is actually the search backend;
You can monetize this by creating a searchcoin reserve and selling it on future, or by receiving searchcoins/safecoins for each query, or by selling ads, or doing it for free.
You could have a feedback grading system or something. Level 1 is cheap but the feedback isn’t that in depth. Level 2 is better but more expensive. Level 3 would require users to take even more time to provide feedback, thus get paid more, thus be even more expensive. So on and so on. As users desired more comprehensive analysis or degrees of work the price could go up. Or you might just make it modular and stack functions on top of one another. List a url? That’s one job. Check to see if there’s any banners on the homepage? That’s another job? Look through the entire site for adverts/banners? That’s a slightly bigger job. Tag sites that solicit information or ask for payment? That’s another job. Etc, etc. How does one define spam? Is Mozilla soliciting their mailing list at you considered spam? Lot’s of tech groups ask you to join their mailing list. Personally I can find that a bit annoying but I understand where they’re coming from given that a lot of tech groups prefer mailing lists as their preferred mode of communication. But email is also used in marketing. So being asked for one’s email can be considered spam. Hence how does one define spam? And how is it presented?
I understand the value of two factor authentification but I also understand how one would NOT want to give a corporation their phone number. And I understand how annoying it could be to be constantly pestered about such things. This is why I think the whole stacking of functions might be a good approach because it allows people to assert their own values when assembling their “anti-spam” package.
No ads, no sponsor influence, no selling people’s information. It must be fully decentralized at some point in some way. Use AI. It doesn’t have to make money and probably best if it doesn’t it could be pure FOSS community built and driven a little like Wikipaedia but better on SAFE.
Consider the Etsay model. Provider only gets paid when a buyer is matched to a seller. Again no ads because accurate search totally obviates ads ads violate trust and are evidence of the worst sort of conflict of interest. Again search only gets micropayment or cut when a buyer and seller are actually matched but carries general unsponsored accurate indext as good will and necessary attractant to service. Service never favors buyer over seller. Kind of a Craig’s list model. It never takes money from either does free listings, only gets paid when it accurate matches buyers to sellers. Assiduously avoids contaminating general search query with commercial search result keeps it honest to keep good will and efficacy high. Has a no BS company behind it uses flat cooperative model doesn’t do stupid things like investing or investing retirment funds- only takes money from legitimate end customers for services honest services rendered,. Likely gets rid of supidites like boards, execs, managers and supervisors- is probably close to pure DAO or is pure DAO.
Search is the heart of everything its one thing not to F up. It must fully decentralize even if that waits for the disributed computing function. I remember asking if search would be ready at launch, but this is MVP.
And the AI would be programmed by Google.
Okay, maybe not, but you get the idea.
This type of reasoning - that AI may be the answer - leads some to believe that because people are selfish and greedy, electing the right government may put good people in command of the bad. But invariably, given monopoly on the use of force, central banking, etc., the new people soon turn out to be selfish and greedy. Or they get assassinated, as the case may be (it’s hard to say what can be done in this case, except that the good guys, as soon as they come to power, dissolve the State apparatus as soon as possible).
Whoever can make more money from search engine he creates, will have more money to invest in its development, so ultimately there will be a SAFE google.
Yes but Linux over Windows in this case. But I do see. Try to replace unacceptable human govenrment with human made code and choose the self modifying variety and end up with an increasingly unintelligble universe with an increasing number of physical laws that are increasingly unintelligble. Chocolate ice cream becomes impossible but vanilla still works.
A comment from David regarding development of search in SAFE:
Search though from a decentralised perspective can make use of lucene type indexing, with the indexes using using Immutable Data, this has to be arranged in a manner where it can append information as more links and data is found.As data can be referenced via a datamap hash then the end data will always be found (like way back machine inbuilt to the network).
So the research part if you like will be fitting this in a type of binary tree/Directed Graph where leaves can be updated to reflect new information that fits that part of the object (graph).
So very possible and a few whiteboard sessions could product a crude first attempt at this. If done correctly and without doubt in the open then I think we can have a decent, albeit crude search.
More interesting though will be the use of deep learning algorithms to provide more than search, so more like wolfram type question engines with larger data sets. I spoke briefly with Wolfram and I am sure there is a very long conversation to be had there, with the new data structures available in SAFE as well as their capabilities in domain specific languages in this field.
I wonder, given ‘human’ indexing, and the problem of Proof of Unique Human, if spam could be alleviated by the first (or maybe every rating) incurring a fee in safecoin? That could/would hopefully prevent mass spam scripts if the cost was high enough.
Given that, the problem would be finding this sweet spot with the search engine; though I imagine if it is human powered / spam free then it would be worth the initial outlay for users who value private search etc.
This cost could be redistributed to the search site’s indexer team… Mo rep mo money. Take some cut for the dev (open transparent amount). Boom? Everyone wins.
Pay to submit a link to the engine. If your reviews are good and valid, youd garner that back over time. Perhaps even make koney off of it. But would stop account creation and just ‘reviewing’ your spam site.
Ahh OK I see what your saying but I doubt it would work. People have to buy domains and hosting currently, doesn’t stop them from spamming also on SAFE net they will pay to PUT content anyway.
