joshenders

> - A search engine is good when it provides results that are useful and what you wanted.

Agree and I would add that the many relevant results you see today would not appear if it were not for Pinterest users adding context to images and the engineering work to surface these pages and exposing context in a way for Google to fairly rank.

> - Seeing pinterest links in google search is not useful and not what people want (just read this thread) > > Therefore, by making pinterest results dominate google search, you have made google search worse. > > Google search is the main way people navigate the internet, so by making google search worse, you have made the internet worse.

There’s a lot to unpack here so let’s just start with a simple question: I would ask, “worse” as measured by what?

Pinterest has staffed a team of dozens for almost 10 years to improve incoming traffic from search engines. Their effectiveness is measured by week over week increase in traffic and sustained use of the product from those users.

Literally 10s of millions of increased MAUs from this one channel alone. For these users, they are clearly finding value in the content that is being surfaced. Google also clearly sees value in surfacing these results over others. If you think Pinterest has somehow been pulling the sheepskin over Google’s eyes for over a decade, you’re mistaken.

I’ve worked for companies where I wasn’t the target demographic and I had to constantly remind myself that just because I wasn’t the target audience, that does not preclude a majority from realizing immense value.

I was responding specifically to the notion that Google should block all paywall or signup gated sites.

I see your point and I think you understand the product better than most in this thread but I think you’re minimizing the value of enrichment that “pins” over images or static links provide. In the same way, the comment section enriches news articles on hackernews and this site is worthy of indexing by Google.

Hyperbole aside, can you describe specifically what about the user experience you find abysmal?

I understand the frustration with speed bumps but for the majority of people (US users at least) the signup nag is dismissible and for the rest of the world, it’s a simple speed bump which helps maintain the operation of the business and provides real value back to the end user.

I also have a Wikipedia account, a Stack Exchange account, a YouTube account, a HN account, and countless one-off forum accounts. This is just the reality of the modern web and per-user customization.

Do you consider hackernews to be “hijacking” news content?

Pinterest is no different.

I think there’s some confusion with how the product works:

The images are scraped on behalf of user request along with relevant information from the originating webpage. The goal is that all pins on Pinterest are actionable (in nearly all cases, linking to the original webpage). The most often cause of missing links is for a website to go offline. In those cases, would you rather your Google image search return no results or for a cached image on Pinterest with a link to the original website which can be used for further digging.

> (probably violating the copyright holder’s rights in the process)

No, that’s not what is happening at all here.

I was able to repro with a VPN on .co.uk . It looks like for some ccTLD (target markets) the nag UI isn’t directly dismissible. This seems to have happened after my time there. I understand the frustration. Either create an account, use a US VPN, or redirect yourself to .com with a specific cookie value set. I’ll leave that exercise to you.

Here are some test URLs which should all be publicly accessible without an account:

User page: https://www.pinterest.com/joshenders/

Board page: https://www.pinterest.com/joshenders/herpetology/

Pin page: https://www.pinterest.com/pin/46232333644621276/

By “expert user” I just mean “power user”. Didn’t intend for that to come across as an insult.

Scroll a little bit, the X is in the top left. We made that change in 2017 or so.

Just confirmed it still behaves this way on .com for keyword landing pages and direct links to pins (e.g search results, or shares from users).

*Pin quality, not relevance

Disagree. If you want to debate me, let’s hear your argument.

If “this” is the signup nag, then yes. It’s largely ignored because the nag can be easily closed and only non-expert users don’t understand this.

I don’t think you’d be very satisfied with your remaining search results.

After answering some other questions it occurred to me that Google may have been incentivized by our intention and ability to improve results for image searches. E.g at scrape time, when an image becomes a “pin”, we were essentially creating a enriched view of that image for which Google is able to surface.

Imagine a scenario where the original image was linked in the page as “main.jpg” or “flower.jpg”.

There may or may not be useful signals on the page for Google to infer from. In the case where the image is saved on Pinterest, the title may come from an expert user who submits, “Trumpet Daffodil”. This is a lot more precise and searchable.

We would go out of our way to append keywords to the end of page and image filenames to make those even more relevant. A final file name for this image would be, “eb761df34b9219ffc68e5c9e55a134f0—-trumpet-daffodil.jpg”

I’ve been exposed to data on this for years and I can say concretely that the average user is getting more relevant image search results than ever before… many just happen to be “search optimized” and “proxied” by the same punching bag, Pinterest.

Sign up is free but if you’d rather not, just close the sign up nag.

That’s an interesting perspective but I don’t think it’s well calibrated for the modern web.

You’re mischaracterizing the user experience. As a logged out user the signup nag can be easily closed or ignored and is identical to the logged in user’s. Often times Pinterest has a copy of the image and metadata when the original site may be offline.

The effort behind search engine optimization wasn’t based on the goal of improving any specific product flow (usually). Most if not all changes were gated behind the outcome of before/after and A/B experiments which were almost solely measured by growth and increased traffic. The problem is that there is obviously a huge disconnect between activity and experience. I often made jokes about putting a 100% interstitial “click me” div over the page and demonstrating our impressive click through rate as a win.

So I would say, the incentive here was to increase search traffic and instigate user activation. Many folks now work at Reddit and are repeating the same playbook. Many folks cut their teeth at places like Facebook or travel search engines.

I was at Pinterest from 2014 to 2020, so a total of 6 years. I lead a team and was enamored with the product. I saw hyper growth from a 250 employee company to 3,000. I rang the IPO bell at NYSE.

Was Pinterest shady? No, not at all. Most employees were kind and genuine and trying to do the right thing. Was management and leadership all hyper corporate lizard people? Yes, absolutely. Was it a good company to work for as an engineer? I would say no.

The company constantly suffered from lack of quality technical leadership and as a result the lizard people were in control and the politics were of a company 10x at all times. Decisions were almost exclusively top down. Sound technical ideas were difficult to argue for because shady employees would lie and manipulate the lizard people. The lizard people only cared about self preservation and promotion.

I also think Ben and Evan were heavily influenced by their board (Ron Conway, et all) who saw a billion dollar company and wanted to not rock the boat until exit. It could have been so much more than a cash grab and yet I don’t feel like it ever was or ever will be.

Please elaborate

Yes, absolutely. I would love nothing more than for every Pin on Pinterest to link to the relevant website or have rich contextual information. I honestly don’t think it was that difficult technically (in most cases) but IMHO “pin relevance” was underinvested in at Pinterest. Ben (CEO) and the leadership team would often make public pleas about “how important it is to increase pin relevancy” and I saw very little action in the engineering organization to support that.

> You have us completely irrelevant hits.

Pinterest is intentionally designed to facilitate rich context around images. That’s kind of the whole idea and value proposition of the service.

Most “Pins” are created with the Pin-it button browser extension or JavaScript embed which associates the website the image is on and other context with that image and retained when “uploaded” to Pinterest.

In the long tail, yes- there are Pins that are void of context because of incorrect user submission (direct image upload) or broken links as a result of the website owner not redirecting when the content is moved. This was a continuous problem that everybody was interested in solving.

Sure, just as any other site is.

The source in the article speaking about “size savings” of reducing images is clearly non-technical. That was not the reason as our S3 and CDN budget was essentially infinite.

I would say the main negative of Pinterest in image search results is when the target of your search isn’t the image but context around the image. It’s often lost if the image is reshared from an unrelated site or directly uploaded to Pinterest without relevant metadata (like the site it came from). I think a lot of people forget that GIS often gave contextually out of place images even before Pinterest but now there’s a consistent domain to associate with the frustration of missing context.

No. We absolutely did not pay for ranking or anything like that.

I don’t have a good answer for that but I can describe the feeling I got in these meetings:

They were pretty guarded and cagey about everything to be honest. Felt like a favor from a trusted internal contact but our contacts there would not share any details and mostly listen and occasionally confirm our guesses if it was public knowledge. Our meeting agendas were always prepared well in advance and very intentional.

Hi, I worked at Pinterest and heavily supported the “Growth SEO” team from the infrastructure side 2014-2018 including attending weekly joint meetings between Pinterest and Google. I am primarily responsible for many key internal initiatives that increased relevancy of our images in Google Image Search. AMA

Flag the article. I noticed this too.

jimsi, employees are also extremely active on Twitter. It’s not the most scalable but that’s usually my goto for eyebrow raising Cf related questions.

Good luck constructing additional pylons

Reddit siphoned Pinterest’s growth team. It’s no surprise.

>> For sites of any complexity with any dynamic content having CDN redundancy is akin to being multi-cloud — it is not worth the effort.

I proposed and lead our multi-CDN project at Pinterest for both static and dynamic content and I can tell you, many many times over, it has been well worth the effort. Everybody should do this if not only for contract negotiating leverage.

Cache invalidation is fast enough on all CDNs now for most use cases (yes, including Akamai). But realistically, most sites (Pinterest included) are not using clever cache invalidation for dynamic content because it’s not worth the integration effort (and it’s very difficult to abstract for large 1k+ engineering teams). Most customers are just using DSAs for the L4/L5 benefits (both security and perf). In that case, it’s not complicated to implement multi-cdn.