The problem is that a tech company like Google is incredibly valuable for a simple reason- their marginal cost is ~$0. They have a fixed cost to run the service, relatively tiny marginal costs and massively higher marginal revenue so it scales fantastically.

If you want a curated recommendation site then your marginal cost is going to be high - every geographic location you want to be in you're going to need to find every single business and invest significant time to investigate them. That time investigating all the coffee shops in Boston isn't going to help you win customers in Minnesota. That's a perfectly fine business, but it's not going to scale well and when you do finally review every coffee shop in Minnesota, Google is going to crawl all the data on your website and stick a card on its search results with all the information killing your site overnight. This has happened before.

Because the high-value method doesn't have a nice business model, the companies that attempt it all either die due to being ripped off, being unable to scale, or get seduced by the easy growth of automated & user-generated content.

The scaling is when you are big enough that businesses want to be listed on your site. Then, rather than letting them put up whatever, you start curating submissions.

If you can break the maps moat of 'everyone uses it' then it doesn't matter if the marginals of maps are better, because people won't use the inferior product.

Hopefully. I'd happily pay someone for curated listings and reviews (we don't seem to have something like Consumer Reports over in the EU; at least I don't know of anything). Everything that's not either a recommendation from a person you know or community forums around a particular niche seems to be thoroughly gamed.

There's nothing EU-wide, but many countries have their own versions. In Sweden there's "Råd & Rön" which was previously published by the government consumer advocacy agency and now by an independent Consumer's Association.

See this link for a list of EU national organizations https://en.wikipedia.org/wiki/The_European_Consumer_Organisa...

Juicing engagement necessarily comes at the expense of trustworthiness.

(I initially inadvertently typoed "truthworthiness", which I think is more apt.)

There's the Stiftung Warentest in Germany (test.de).

Which? magazine in the UK.

Just curious - what's wrong with Angie's List? It seems a little harder to create an account (they actually did a small background check on me and called me on the phone to ask some questions when I signed up - it was kind of creepy), so it seems to cut down on the random postings from paid post farms and fake reviews you get on places like Amazon, Yelp, etc.

Like Yelp, Angie's List businesses can pay for more prominent placement, and AL uses dark UI patterns to make it seem like they're somehow more legitimate than other reviewers, when in fact they just got there due to paying more. Same with Yelp.

One of the worst experiences I've ever had with a contractor was with an electrician who had 117 A+ reviews on Angie's List (K&K Electric of Rockville, MD). They took 7 hours to complete a 1.5 hour job (installing a 40 foot cable run for a car charger), then threatened to put a lien on my house because I wanted to pay less as a result of the experience. It was a total shouting match (full disclosure: part of the tension was that I had to pay them 40 days later since I was out of the country for 4 weeks). On top of all that, the permit was rejected because they didn't properly label the high voltage, high amp breakers.

Now I realize they probably paid for those "117 A+" reviews.

There's lots of great info on Yelp and Google reviews, if you know how to filter it. Two distinct filters are required, actually, one for the corrupt, and another for the clueless.

Corrupt reviewer are the fake business locations, paid shills, employees, relatives, customers trading a 5-star review for a 10% discount. Smoke these out with the following tells. Review history: absent or all 5-star. Text content: description of service absent or generic description of service, zero negative or neutral specifics, hyperbolic declarations of loyalty.

Clueless reviewers leave easier tells, but can be more plentiful. Review history is bipolar, all single- or five-star. Text content shows misunderstanding of what the vendor actually does, expressed with many errors of grammar, and irrelevant detail. May be excessively negative, due to unreasonable expectations of the vendor.

So, reading a good sample of reviewer's text is required. This does consume a few more minutes of your day.

Were the Yellow Pages ever actually curated? It's been a very long time now, but back in high school I did some freelance computer repair work by making up a business name and putting it in the Yellow Pages with my cell phone number. As far as I can recall, there was no validation of anything other than that the check I wrote them was good.

The curation for yellow pages was more indirect than anything.

A) your check cleared

B) you managed to work with the yellow pages company to get your ad right

C) you managed to keep your phone number operating between the time you placed the ad, and the time I opened the book to use it -- probably at least a couple of months for layout and printing, plus the up to a year before the next printing.

That’s validation in of itself. Google does this with Chrome Apps — you pay $5 to publish.

> User-fed databases like this are all full of absolute garbage data, and have been gamed.

Is OSM equally bad? There's many ways to run a "user-fed" project.

OSM isn't, but I suspect it's mostly because it's not high profile enough (the old "nobody bothers writing viruses for Linux" problem). If it were, I'm sure we'd see junk data. OSM's data is version controlled in some way, so I'd assume it could be rolled back, but that sounds labor-intensive.

I use OSMand instead of Google Maps, and love it, except for the absolutely atrocious address parsing when searching. There's also a general lack of street numbers in my area (Pacific NW USA), which can make things difficult.

Amusingly, it also suffers from the "won't show you street name labels" problem that other posters are accusing Google Maps of.

Using the app Streetcomplete, it's really easy to correct OSM data like street numbers and opening times around you as you walk.

I have been working on adding all street numbers for a small town of 40,000 people. Two years into this project, I am still little more than halfway done. Adding a handful of street numbers with StreetComplete might be "easy", but it is incredibly labor-intensive to add street numbers in an amount that makes OSM actually useful for navigation purposes. Google only managed to get their house numbers by forcing the world to do the work through ReCAPTCHA or by buying the rights to various preexisting databases.

I wonder if OSMand could take advantage of having a few street numbers. Right now, in the "address search" mode, if the number you're looking for doesn't exist, you're out of luck. But theoretically it could accept any street number, and if a street even has a few numbers in place (I often find random street numbers probably entered by residents or shopowners) it could at least tell you "it's somewhere between here and there". Even that would be vastly more useful.

Then mapped crusaders like yourself could first focus on adding numbers in the middle of streets, and then gradually, in binary search fashion, continuing to halve unnumbered street sections.

One of the biggest challenges with mapping house numbers is simply getting to that street to begin with. It takes a volunteer making an effort to go to various parts of town that he/she doedn't normally go to, since most neighbourhoods don't have their own local mappers interested in adding house numbers.

Once someone has made that effort, then it isn't too much extra effort to do all the house numbers. So, no need to restrict oneself to the middle of the range.

OSMAnd's address search feature already allows you to choose from among the house numbers that do exist for a street, even if the whole street isn't mapped, so it is often easy to navigate to the general ballpark of your destination.

If you can make it a as addicting as pokémon go, it might work. It looks like they are trying since you get some points for "playing"

OSM's Slack is full of people patrolling spammy SEO edits and stuff. It's swiftly dealt with.

If the problem were much larger, it might overwhelm the people currently cleaning it up, but if OSM were higher profile, it would also presumably attract more volunteers.

Crucially, it would empower the people wronged by spam to do something about it, which is precisely the opposite of what Google's system does.

OSM has a community in most places. Nobody really cares about google.

A few years ago some idiot marked up pretty much every walk way in my town as a cycle way. As a keen cyclist who lives in a town with a somewhat poor provision for cycling, this is somewhat irritating as it makes our local government look like it is doing well.

I followed the feedback link and reported a whole heap of these, including links to Streetview showing the "No Cycling" signs in some places. I got automated replies to say that someone would deal with my report - but nothing happended. They are all still showing up.

Perhaps even Google doesn't care about Google.

There was a situation that happened in Thailand, where Grab (a ride-sharing company), outsourced OSM edits, and the company making the changes was doing via satellite pictures, and opposed to ground-truthing.

See https://techcrunch.com/2018/12/19/grab-maps-osm-thailand-sou...

No, but mostly because nobody directly uses OSM. They all use Google.

Same with Wikipedia. Wikipedia has been gamed a bit but the excellent moderators have kept many things at bay. I don't think it's impossible for someone like Encyclopedia Brittanica (an online version) to resurface.

Market share has something to do with it too. If as many people used OSM as use Google Maps, I'm sure that it would become just as bad.

I wonder if China has a social credit score API that companies can integrate to prevent sign-ups from known bad actors. Would be an interesting concept.

ReCAPTCHA v3?