I really love the vibe of Koh Lanta in Thailand. It has tourist amenities without really being touristy. I find it charming, slow and cheap. Is there a place in Croatia like it?
Completion:
There is no exact place in Croatia that has the same vibe as Koh Lanta in Thailand. However, there are places in Croatia that have a similar charm and slower pace. These places include Hvar Island, Korcula Island, and Istria. While these places do have some tourist amenities, they are not as touristy as Koh Lanta and still offer a more authentic Croatian experience.
Another:
There is no exact equivalent to Koh Lanta in Croatia, but the small town of Nin on the Dalmatian coast comes close. Nin is a charming, slow-paced town with beautiful beaches and a relaxed atmosphere. It is also quite inexpensive by Croatian standards.
It will require multiple dozen auxillary Google searches and reading through countless articles and message boards.
GPT-3 is a great research tool as a brainstorming engine, but sadly cannot currently be formally trusted to output truthful information, even with conditioning.
It takes someone with prior knowledge or subsequent research in order to make use of the result. I'd say this isn't what the author was intending.
Humans, unless they're disturbed or malicious, tend not to hallucinate things out of whole cloth.
Me: What is your name? Model: I'm Mark A. <continue> Dreyer, and I'm with the <continue> Federal Communications Commission (FCC)
Obviously, it's not actually a guy named Mark A. Dreyer and, as far as I can tell, there is no Mark. A. Dreyer in a prominent role at the FCC. That, as far as I'm concerned, is essentially a hallucination.
If you think about how they work, this isn't totally surprising. The model has been programmed to take the input, turn it into a series of tokens, and then produce another series of tokens that are especially likely to follow the input ones. That's it.
To the extent that this corresponds to reality, it's only because reality is overrepresented in the training data, and so token sequences corresponding to reality are often more likely under its model. This also explains how it goes off the rails: Mark is a very common name (in the top 25 for men for decades) and once you generate a stereotypical middle-aged white guy name, it's not a huge leap to him being a bureaucrat, etc. I forgot to delete the huggingface boilerplate, which talks about advances in technology (etc), which probably sent it down the FCC route too.
Something like coming to accept some data-set local optima as a global truth and deriving beliefs without underlying basis.
The whole point of these models is to make up something that seems "realistic", but not regurgitate the original contents. If that's not hallucination I don't know what is.
Prompt: "Is Fortigate open source?"
Article: "Based on the parameters we compared, it seems that pfSense is the more favorable solution because it is open source and also offers great features."
Google: "It is open source!"
query: what is the koh lanta of croatia?
response:
Koh Lanta is an island in Thailand that's known for its beautiful beaches and clear waters. It's a popular destination for tourists, and it's also become a popular spot for digital nomads in recent years.
Croatia doesn't have an island that's quite like Koh Lanta, but there are a few options that come close. The island of Hvar is a good option, as it has beautiful beaches and a laid-back vibe. The island of Korcula is another option, and it's known for its medieval villages and stunning scenery.
Anecdote: When I told a random person I got there by plane he automatically assumed I owned a plane."
I wonder how long it will take before we have a GPT-3 bot commenting here on HN
Needless to say, all of these people get told to fuck off and get a job.
https://www.theverge.com/2020/8/16/21371049/gpt3-hacker-news...
All the results it gave are flooded with tourists, and some of them are among the most expensive you can get in Croatia. :shrug:
Unsure of what to think of the results, as it seems to have very little relevance to the input description other than being in Croatia and being tourist spots.
I could write pages of this stuff, but let me write more positive things. The best way to enjoy the islands in Croatia is actually by boat, rented or otherwise (they are not expensive). There are beautiful quiet coves where you can spend your days and nights, only going for resupply in towns. You anchor in a quiet cove, and in the evening you take your tender boat to the town if you want to walk around it. Next day you sail to another island and repeat. That experience is actually awesome.
If a boat is out of the question, being in some apartment is an option, I would just advise not to be in the town center or near it, or anywhere where some very local beach is. Mostly there will be a caffee with blaring music and nobody likes to sleep with earplugs. Also this is and advice for any island, and true, a lot of those islands are quite different than each other.
This shows, if you had any doubt about it, that it did not actually understand the question.
What we get instead are incorrect results which are being presented as reasonable answers to the search query in the name of advertising revenue.
When I discovered Google (I think it was running at beta.stanford.edu at the time) it was just short of a miracle.
There is no blank space to fill with something useful.
Instead everything is filled with SEO driven, ad riddled, bullshit.
And creating something useful is more cost and effort than creating more SEO driven crap.
And even if you do create something useful, you’d then need to focus on SEO in order to compete.
There’s no way this problem isn’t solvable. But google seems to lack any motivation to solve it.
Ugh, this reminds me of the people that move from high tax places to low tax places but want all the benefits of high tax places at low cost.
Now, back to the main point, if this was searchable it wouldn't exist.
The number one complaint people have today is popular places are too busy, too many tourists come and prices go up and it's over ran. So what do you think is going to happen to some place that 300+ million wealthy people around the world can search in a few seconds and find out it's cheap an empty. The answer is "become busy and expensive".
The only reason places like this still exist is Google can't find them.
Here's a tip for the blog author. Learn a foreign language. Then search google in that foreign language. You might be surprised. Be the change you want to see.
Youtube is worse, I have two YT accounts in English where I only watch videos in English, and Youtube keeps recommending me videos in French that are not relevant at all to my interests (even though most recommendations in English are perfectly relevant). It's as if their AI just saw my IP and decided to recommend to me the most popular French videos of the moment.
What I needed was to use a different search engine. DuckDuckGo gives me a toggle to include or not French results.
I passed back through a few months later and barely recognised the place. Wooden chalets has given way to hotel rooms and beachside stalls to night clubs.
The whole point of even having tourist amenities is that there's a community that that's attempting to draw tourist money to their location, and probably create a thriving economy.
An unknown tourist attraction with amenities is a failed location where the amenities are unmaintained/dilapidated, the local economy has nosedived, the young have moved out and the population is collapsing.
> there's a community that that's attempting to draw tourist money to their location, and probably create a thriving economy
Usually that's not how it works (at least from first-hand experience). It's about a local business/mafia trying to draw tourist money while exploiting cheap labor from the local community. Let's not even get started on the scheming to privatize public beaches, environmental pollution (noise, lights, trash) and other shenanigans the tourist industry is creating to the detriment of local communities.
In the context of the discussion, I don't understand your question. The money comes from the hypothetical tourists who're hypothetically making all the usual spots too busy. What point are you making?
It was less likely to ignore the absence of a particular keyword but certainly wasn't better at NLP
I don't think Google of 15 years ago will do a good job on the current age of internet. The size of web has increased a lot like 100x where most of the increment comes from unstructured formats like video or image and its signal to noise ratio deteriorated considerably. SEO has become much more sophisticated ever. A large fraction of useful information is now locked in the unindexed walled garden.
In fact it could be less than 1x in many cases. People that would have blogged about their Croatia trip might now instead just have a 10 word tweet, an instagram story (of similar low quality and probably not publicly indexable), or a youtube video (so non text content)
Well, I was about to post a similar reply. It would have taken some messing around with the search string, but I'm confident I would be able to find something like this on Google 18 years ago, and I'm certain that if it exists, I can't find it now (unless it's on some big name site).
And then you learn that on a good day you could see Croatia from Venice, or something like that.
(I just completely made that up, don't quote me on that.)
Their apparently paying a lot less attention to certain things (like link text—"what might someone use to describe a link to this resource, but which might not appear on the resource itself?" used to be a very fruitful way to search), and freely substituting words or dropping terms that are merely not common on hits (but not totally absent from 100% of results) has made this kind of thing impossible.
That being said, I remember my cousin had a blog where he was writing utter nonsense about trending topics and he was actually able to pull enough money from Ad Sense to afford a pub crawl once a month.
Now of course all of this is streamlined - one button website generators with AI content and posts for given topic and self optimising for engagement and advert clicks etc.
But I see Google has thrown in the towel and no longer cares for search results.
The reason even a fairly advance NLP system is going to struggle to find what the author considers to be the "Ko Lanta of Croatia" is that they can't read the author's mind about qualities associated with Ko Lanta, so they're going to highlight islands, not "small, quiet resort villages with $10-$20 rooms and hammocks" (Not really Croatia's "vibe" though I suspect the OP would probably like most Croatian islands anyway...). The same is going to be true even if you put in strings that return comparisons humans directly make themselves like "Venice of Asia".
Once I lived in the city, it was too big and noisy
So I moved to the country to stop and smell the rosies
All my city friends joined me and put up nice new housies
Now it's too big and noisy, think I'll move to the country
https://en.wikivoyage.org/wiki/Previously_Off_the_beaten_pat...
Like, Koh Lanta is surely not just a vacation destination, it's also a place where people live their lives, go to work and school, are homeless, deal with illness. The expectation that Google will assume that by "koh lanta of croatia" you're really asking "vacation destinations in Croatia that are similar to Koh Lanta, Thailand" really rubs me the wrong way
> Places where things move at a less hurried pace, where Croatian life can be savoured, where you get a flavour of what the Dalmatians call fjaka – the art of doing nothing. These islands and mainland destinations are what you want in a post-lockdown escape: peace, beauty and the chance to discover why Croatia is such an enticing country.
Now I know I want to search for fjaka -- if I want to search further because this article might just be what OP wanted.
The success for search always has been finding the keywords to search for. You can see here it took me just a few minutes.
Fed a bunch of countries coastline into it, and popped out ranked beachfronts to investigate more.
1. I once went to Ao Nang, Thailand a few years after the tsunami and found a really cool chill cheap beach town and loved it. Went back a few years later and right on the main street there was a McDonalds. Across from that there was a Burger king, ~100M up the road was another McDonalds. I will never return.
Where did you get the data from? Do you still have the code somewhere?
It worked for a select group of toy problem domains and the answer for most high-concept searches would be noise or nothing.
The author never manages to convince me that the information they're looking for is on the Internet for Google to index in the first place. The author then asserts "don't just hit me with garbage," which is a fine assertion, except what they're interpreting as 'garbage' is the information other people making the search could actually use. Google A/B tested the hell out of "say nothing" vs. "guess something close to what the requester might care about" and the latter won out every time.
If this data is there now, it is most likely in a Facebook post which Google isn't seeing.
Google could find them on Archive but we both know that's not the business.
As is every query that we need computers for (e.g., a SELECT query that goes over millions of records).
On the other hand, expecting a search engine to find pages which at least mention both Croatia and Koh Lanta is a very, very, very low bar.
And it's going to be better than what Google is giving today.
It sounds like this person wants to have a philosophical discussion with a friend about this type of subjective question rather than simply receive a factual search engine response.
Tried and failed.
How would that prevent spam?
What seems like an endless army of Google proponents (perhaps they are simply "status quo" proponents) discourage anyone from even attempting to think about how the web could be organised without the need for a Google.
In a library, I can do searches restricted to certain subject areas. I might only be searching certain databases that pertain to certain subjects.
Whether it could ever be feasible to search the www by "subject area" is left as a question for the reader. In any event, if it were I would search for "koh lanta" in a relevant subject area, e.g., "travel reviews". One can search for pages that contain "koh lanta" on www sites using domain names registered from the .hr registry, e.g.,
"koh lanta" site:.hr
Even if I could just obtain a list of travel review websites, I could index those sites and then search that index for reviews that contain the terms (a) "croatia", or names of various destinations in Croatia, and (b) "koh lanta". If I was really patient I could use a script that simply performed the same search on each site, either using the site's "search" option or a search engine.
I am still not sure this search even makes sense. It seems to rely on an assumption that some traveller will compare Koh Lanta with some location in Croatia. What could make more sense is to define what are the specific charactersitics of Koh Lanta that one wants to find in a Croatian destination, then search for those. Then I might search not only travel reviews but sites that describe characteristics of Croatian travel destinations.
In any event, it is this sort of catgorisation of sites that is generally missing from the www. I believe it is feasible but "tech" companies like Google and its followers are not interested in promoting such facilities, preferring instead to pursue data collection about www users and programmatic online advertising. If Google is the "front page" of the www, then it can portray the www in the way it sees most beneficial to Google. For example, an endless sea of disorganised information that is impossible to utilise without Google's assistance. I certainly do not need to search billions of web pages to find the "koh lanta of croatia". I only need to search sites with travel reviews. But I do not get to limit a search like that. An advertising company gets to decide what is "relevant" to the terms I input. _Popularity_ (potential value for advertising to a wide audience) dictates what I can and cannot see of the www through Google.
About 10 years ago I started a proof of concept that aimed to catgorise a web (cf. "the" web) via non-ICANN issued domain names issued by a non-commercial domain name registry not governed by ICANN. The idea is that the FQDN can contain 1. a subject matter description (subdomain) 2. a trademark (domain) and 3. a Nice trademark class (TLD). Searches can then be done largely based on FQDN instead of heuristics such as "page rank", popularity or other metrics designed for advertising purposes. This new web has value to me because it is constructed from legitimate organisations that have invested in trademarks, i.e., they have a registered address and they pay their legal bills. Registering a mark suggests they have an existing or future brand to protect, a business that needs a mark to protect consumers from becoming confused about the origin of a product or service. (How many of those Chinese sellers on Amazon today have trademarks.) Ideally, this system filters out those who are not legitimate businesses. The garbage at the top of Google would not be possible. To game this system would require obtaining trademarks, not simply gaming ICANN's system of registries and registrars. (And we know ICANN itself is not a trusted steward of DNS, but a means for a select few to make huge profits from it.)
This prototype web was not meant as a replacement for "the" web, but as an experiment to separate out the commercial entities on the www from the non-commercial ones. Google and its ilk want a www where there is no distinction between commercial and non-commercial www use. All www use is surveilled for commercial purposes.
This mixing of commercial with non-commercial, to me poses one of the biggest threats to the www we started using in the early 1990s being wiped out. In the early days, there was this idea that TLDs would represent different categories of websites. For example, ".org" would be non-commercial websites, ".edu" would be educational institutions, and so on. Things have changed. The root.zone has exploded in size with countless "gTLDs", most of them are purely commercial. History is showing that it is infeasible to demand that TLDs enforce some sort of rules over the contents of websites that use them. One needs "Google" to help figure out what websites are worth visiting.
The prototype web in theory allows one to find legitimate brands on the www by searching registered marks or subject matter descriptions within Nice trademark classes that are contained in FQDNs, not the contents of web pages. It is the antithesis of what one sees on Amazon today, what with a gazillion Chinese knock-offs, including Amazon's own "brands". There is no "SEO" on this new web because the contents of web pages, e.g., "backlinks", are not the basis for search.
But do not worry, the prototype web does not exist and would never work. It is a terrible idea. Google will never die. See you in the Metaverse.
Well exactly. If anyone ever mentioned Koh Lanta in their review of Croatian destinations, we'd hope that that it would be at the top of the results.
Or that there would be no results.
Getting results that don't even mention Croatia is not the intended outcome.
Also which country's trademark office would be the gate keeper? There's one in every country. I assume you mean the USA trademark database in this case, but that only covers US trademarks.
This is irrelevant to the purpose of the "prototype web". It is a different problem. The purpose of the prototype web is not "to make things much more difficult" for "spammers". Its purpose is to _separate_ the commercial web from the non-commercial web and to _categorise_ the commercial web in a way that makes web search easier.
Given that trademarks are allegedly "easy and cheap to get", akrymski could share with us how many he has registered. Surely it would be equal to the number of domain names he has registered since the UDRP favours trademarks registrants and trademark holders have the additional option of using the ACPA. It would make sense to have an "easy and cheap" trademark for each domain name, registered in every class in every territory.^1 How much would that cost.
1. Because domain names have no such limitations.
"Also which country's trademark office would be the gate keeper? There's one in every country. I assume you mean the USA trademark database in this case, but that only covers US trademarks."
It depends on the ISO-3166-1 country code TLD. If the country code is US then the USPTO would be the applicable office.
Forgot about 4. an ISO-3166 country code. Thus, the FQDN becomes 1. subject matter description (subdomain), 2. trademark (subdomain) 3. Nice TM class (domain) and 4. ISO-3166 country code (TLD).
But that isn't the correct result. "Koh Lanta" is being used by the OP to describe the vibe of a place "having tourist amenities without being touristy." It isn't a common English idiom, but one they made up based on their personal experiences as a tourist in Thailand. Results which simply contain the text "Croatia" and "Koh Lanta" aren't going to reflect what they want.
> Google has gotten exceedingly good at organizing everyday life. It reliably gets me news, recipes...
What I mainly hear on this site and elsewhere is people complaining about how the news is partisan rubbish, fake, and serves the "elites". And recipes. Seriously? I had a half dozen people berate me earlier for suggesting that cooking your own meals should be an enjoyable part of life.
Nobody wants news and recipes.
Here's the problem.. "The Internet" was military project that got loose. "The Web" was a solution for research scientists to exchange papers. That's all. Driven by a massively profitable industry, a solution looking for a problem expanded in expectations, and took over the consciousness of generations of an entire society.
"Organising everyday life" is an inadequate appraisal. Most businesses had to be dragged screaming and kicking online. We still possess a vague, poorly thought-through ideal that somehow a connected, "technological" society is a better one, and a thirst for convenience and speed. But let's be honest, we don't really know what the net is for. It has no telos or guiding design principle other than what we overlay on it.
Military tools in the hands of "the people" tend towards insurgency and revolution. Wasn't The Internet's "killer app" the Arab Spring? The brightest moment of the Web was in it's formative years, as an explosion of ordinary speech and new political power. Much since then has been an reaction to try dampening it, domesticating netizens, or even putting the genie back in the bottle.
Maybe a reason to be optimistic, excited even, is that we're still in the infancy of the network. It's still pregnant with unimaginable possibility.
From the article:
> Lets look at some examples. One of my favorite places in the world is Koh Lanta, Thailand. When traveling, I’m always on the lookout for places that give off the Koh Lanta vibe. What does that mean? Hard to say, exactly, but having tourist amenities without being touristy. Charming, slow, cheap. I don’t know exactly; if I did, it’d be easier to find. Anyway, forgetting that Google is bad at long tails, I search for what is the koh lanta of croatia? and get:
This is a near impossible query for human beings, let alone for computer given the state of the AI at this point in time.