Comment by andrewla | Hacker News Reader

Comment by andrewla | Hacker News Reader

Parent Full thread

> Eventually, Google figured out how to create a real moat, by using click data to improve search result ranking

This is revisionism -- Google was far superior to any competing search engine long before Microsoft embarked on its search engine adventures. Google was hand-coding heuristics well into the Bing era. It wasn't until Amit Singhal left Google and search that they pivoted to more machine learning techniques that could use the click data effectively.

This corresponded with the beginning of Google's long decline in search quality, buffeted mostly by the fact that their on-page quality systems were extremely sophisticated at cutting out spam and SEO. The detection was far from perfect, but so many miles ahead of the competitors that their competitors kept unearthing spam that Google had long since excluded from its index but whose fossilized remains still polluted the web.

The moat that Google had was just that they were really good at search quality and PageRank was only a small part of that. In other words, no moat at all, just a better product.

Bah, kids these days. It's not revisionism. I worked on search quality during the hand-coded heuristic era, 2005-2009. I spent some of that time working on the "navboost" team which used click data to alter the search results. Even by 2007, click data was quite valuable, arguably the single most important component of the algorithm.

The algorithm was hand-coded, sure. You don't need machine learning in order to use click data. You just need a few people to have searched for that particular query before. When someone clicks on a result and stays there a while, it's a "long click" and you boost that search result for that query.

Bah, kids these days who think of 2005 as "early times". I had, in one week in July of 2000, during a road show for my startup, meetings with Lycos, Inktomi, Excite, Yahoo, and a few other search engine companies, all of which were considered strong at the time.

Google was already so far ahead in 2000 that everyone who actually cared used it -- although some people still had Altavista bookmarked and used that because they did not know any better. (The internet-using community was much smaller and more knowledgeable back then; Laptops were expensive and heavy and were used mostly by traveling businessmen, the majority had dial-up internet and AOL was the largest ISP. Essentially all internet access was through desktops running Windows 95 or Windows 98. And Google was already head and shoulders above the rest)

By 2005 Google had such an insurmountable edge that there was barely a discussion. Microsoft started development in what would become Bing before that, and despite all their efforts were never really able to catch up to anything that Google was doing.

Yes, click data helps but really only at the head, where Google already excelled. On the tail the increasing dependence on click data (and later on machine learning and an increased focus on user-specific results) has steadily made Google worse and worse, and Google's supremacy has really only lasted this long because of their page quality measures, which is still a huge problem even now on Bing.

Yeah, not revisionism. Google had better search quality for tech users but Altavista, Lycos, Yahoo, AskJeeves, Excite, Inktomi etc had a better results for general users. So tech users adopted Google and when asked by non-techie friends which search engine they used would reply Google. And then Gmail moat appeared and getting a Gmail account was the highest prestige and everyone was trying to get one.

I’ve only started hearing people talk about moats recently, at least by that word.

What happened to building good products?

> only started hearing people talk about moats recently

It’s a decade-old term [1] describing an older concept [2].

> What happened to building good products?

This is necessary, but insufficient on its own. Big “M” marketing is important. (Betamax vs VHS is the canonical example.)

[1] https://en.m.wikipedia.org/wiki/Economic_moat

[2] https://en.m.wikipedia.org/wiki/Porter%27s_five_forces_analy...

No quality of products can compete with a monopoly. The money goes where the largest profits are expected, which is where one hopes to capture a new market

The term moat was used in a leaked Google internal document about the AI space competitiveness, and now everyone is using it like sheep.

It's a Buffett-ism. https://en.wikipedia.org/wiki/Economic_moat

a moat is any product or feature set so good that you absolutely need to use it or it's just that good. for example, Gmail and google maps was a moat for google in the early days.

That's not the usual sense of "moat". A moat is usually a structural or regulatory protection that prevents a superior product from competing without first crossing that moat.

For example, once upon a time building a superior web browser was hard, because the moat you had to cross was that Microsoft would always prefer their own browser, so Windows created a structural barrier to adoption of new browsers. So the amount by which you had to be better than Internet Explorer back in the day was critical, because if you were just a little better you would never get enough adoption. You had to be much better or convince Microsoft to lower the moat.

Bringing up a new stock exchange was another protected industry -- the incumbents had built a huge regulatory apparatus around themselves which meant that to compete with the incumbents, you first had to comply with a lot of regulations which were themselves molded after what the incumbents were doing. Exchanges like BATS took a long time to clear that hurdle, and many exchanges never made it there, even with superior execution technologies, and were consigned to be dark pools instead of lit markets.

Theoretically social networks have had the moat of the "network effect" of having a large user base -- Facebook is useless because all your friends are on MySpace so there's no reason to switch, and MySpace is useless because all your friends are on Friendster. The implications here are that incumbents protected by moats are fragile; once the moat is crossed they can no longer compete because they've languished, so get eaten alive (as the two examples here or say traditional taxi services).

Not quite. A moat is a reason that you can't/won't leave the company/product. An unapproachable feature set might be a moat, but generally a moat is something other than product, that prevents your competitors from succeeding by creating a clone of your product.

I guess a good example of a moat in consumer email could be the fact that once you have the user, their email address is forever tied to your service. They can try to forward emails or fetch them, but they can't retain that email address on your domain without still using you (at least to some extent). Sender authentication then means they likely need to use you to send emails.

That's a fairly effective barrier to switching for many people, as they won't want to lose their email address now, or need to go through every site they ever signed up with and change it.

> I’ve only started hearing people talk about moats recently, at least by that word.

Then you're one of today's Lucky 10,000! The idea of an "economic moat" was popularized by Warren Buffet, starting in 1986. https://xkcd.com/1053/

Can confirm, google was superior from the start. I remember switching from altavista to google and it felt as revolutionary as switching from Windows 3.11 to Windows 95.

Reply on news.ycombinator.com