Show HN: Find recommendations from good people | Hacker News Reader

Show HN: Find recommendations from good people

abyjames.com

256 pointsgits1225124 comments

I always find that recommendation websites suffer in the long-term because the incentive to just keep adding more stuff (to keep the site not seem stagnant, thus make people come back) is greater than the incentive to strictly filter only super high quality things. How is this site going to tackle that problem?

Uncompromising focus on people.

I don't mind if the site is stagnant.

Plan on taking investors? Because that last sentence might make it a challenge, much as I agree with it.

What could investment do for a site like this?

Its a good question.

Scale, curation, combating spammed submissions to AJ, etc. might be the biggest things if the site gains momentum, but if the site remains a hobby project indefinitely (which I think is totally fine regardless of popularity), then it's absolutely not needed.

Doing the right thing makes good money.

Brings to mind "The fisherman and the businessman"

http://paulocoelhoblog.com/2015/09/04/the-fisherman-and-the-...

Haha. Love it.

This is fantastic

If that's true why do externalities exist?

Well just because p -> q doesn't necessarily mean that ~p -> ~q

You have got to be kidding. A list of recommendations is something that is worthy of taking investors?

"Worthy"? Taking investment is not something special, something that needs to be your one and only life mission before you're allowed to do it. If someone thinks an investment to the site will yield a bigger return back, that's enough. Don't let ideas like that prevent you from taking a go at reaching the potential of your idea.

You're right, it (if we're talking about VC) isn't something special. And a highly specific, narrow way of bringing something into being that makes no sense a lot of the time.

Living in San Francisco, you hear just how narrow this focus gets, with 20-somethings walking down the sidewalk making conversation about how to monetize this public accommodation or that human impulse. It strikes me as an odd mixture of funny and repellent, and at least some of those folks will grow up someday.

It does mean a lot of folks are concentrating on a crowded path that smart people will avoid. I mean, sure, if your big idea is in enterprise storage or EV, that's not happening with investment. A recommendation website? If you're not building to flip (which would be stupid anyway), it would be crazy to go begging to build.

I dont use any single recommendation source anymore for this reason. Its always a combination od youtube, amazon recommendations and reviews, reddit, and stackexchange sites.

Great idea. Crowdsourcing is good but averaging out crowdsources of crowdsources is a better way to understand the general sentiment of the population.

Ya theres also alternativeto.net, producthunt, wappylzer, slant for software too. Also bpproductansphoto reviews, bestbuy reviews, and googlereviews.

Then you can also do SEO analytics on site for domain authority too, that tells you roughly how many people use it vs other software. Or check how old a site is, how old a video is etc

I think the problem is that you don’t actually want the general sentiment of the population: it most likely leads to things you’ll accept, but not actually want (ie movies you’re lukewarm on).

I think what recsys really should be trying to target is finding the group most similar to your tastes, and then picking based on the outliers of that group. The goal being to find something within your tastes, but far enough from your normal to be interesting

Recommending based on star personalities preferences, is I think, a way of approaching this idea: why would you want to know what the average item is? You want the best one.

This.

I'll be sourcing recommendations from the silent majority of experts as well (https://prog21.dadgum.com/143.html).

Reference: https://news.ycombinator.com/item?id=17784551.

Im not sure I understand how you intend to do that: the silence problem is that the majority of experts don’t post their recommendations, right? Even if you have their names from linus, how will you get their recs

Email is the first step. I'll have to find another way if that doesn't work :D

OT: I checked your profile but couldn't find an email to contact you. It'd be cool if I can shoot an email once in a while to brainstorm.

There is one area that I can think of, where the recommendation of _some_ 'experts' can very useful: medicine. It is extremely difficult to find a competent physician. Medicine as is generally practiced is often about drastic procedures like surgery, pills etc. which worsens the overall condition of the patient and often results in huge bills. There are various contributing factors, some of which are not in the physician's control - but the end result is always the same - a really bad deal for the patient. 'Silent' recommendations can often lead to some renegade physicians who are great at treating people but may be at odds with the physician's board. All what I said hold true for dentists also: the truly good ones are very few.

Medicine is a tricky area (it is close to religion in this aspect).

Well, agreed...

A few ideas:

- Link the name to a list with all the recommendations by that person, use the date for a link to the place where the recommendation was made (using the date for a permalink is convention).

- Link tags.

- Show everyone who recommended something and sort by how many people recommended it. (For instance The Pragmatic Programmer is recommended by Joel Spolsky, Steve Yegge and Jeff Atwood.)

- Use Amazon Affiliate links to make money.

Thanks for the feedback. I'm building one useful feature a time.

No to amazon affiliates.

I like this. Thanks for sharing.

I hope you will consider one thing, and please don’t mistake it for political correctness. The phrase “from good people” is distasteful and I think we’re all better off not implying such a thing as “bad people”. If the intent is to define a division, it’s important to be specific lest we undermine our egalitarian goals. I know you don’t mean it to be exclusionary, and probably instead aim to evoke positive feelings. But, for me it does the opposite. People accustomed to being excluded are likely to read “from good people” and wonder if that includes them or not.

I agree. I settled on the word good because I wasn't able to find a good word (no pun intended).

I'd use great. Good people is usually used to mean decent people, whereas great people is used to mean impressive people, which may or may not be moral, as in great man theory of history.

Maybe "notable" people?

Consider smile.amazon.com links or affiliates for charities like electronicfro-20 (EFF) though.

Thanks for the suggestion. I will.

> No to amazon affiliates.

Thanks you! This shows character and actual, proper altruistic intentions.

Before reading this, I was not even going to click the link.

Personally, I prefer not to use any project that has no way to make money. Life is expensive and time consuming. Projects require at least a little time and money. If that isn't reimbursed in any way, the rest of life will nearly always take precedent at some point. Affiliate links seem like a good way to solve this. I don't see where the conflict of interest lies.

I wonder at what ROI do we consider an investment something more like philanthropy and at what ROI it is more like capitalism?

Philanthropy is great when it's done by independently wealthy people. If that's who the author of this project is, then that's great and I've misjudged this. But that's not usually the case. The majority of mildly useful little unmonetized projects are not made by independently wealthy folks, they are made by (usually younger) folks who are optimistic and whose expenses and responsibilities have not yet accumulated. I don't mean to be a debbie downer - fun little projects are great - I'm just saying that I don't associate something unmonetized with "great! this is altruistic!", I associate it with "bummer, this will probably not be sustained", and when I come across something with a natural and well-aligned monetization scheme, I feel positively about it and am more likely to use it.

I want myself and others to contribute because of the desire to contribute. I don't want to bring affiliates into the equation lest it brings the wrong incentives (however glimmering it may be).

There are other ways to make money.

I'm honestly curious how you envision that affiliate links might create the wrong incentives. There are other ways to make money, but the ones I can think of seem much worse. Advertising sucks and clearly has more wrong incentive problems. Donations are less predictable and disconnected from the usefulness of the site. It seems to me that getting a small cut of "hey this book does look useful, I think I'll buy it!" is as well-aligned as it gets. But I'm interested in your thoughts on this!

Ads don't suck. Bad ads suck. I love good ads.

I used to wait for ads to appear on my Garena client back in the days when I played DOTA because they were so cool (and useful).

If I know my audience, I don't have to compromise privacy to show useful ads clean and neat either (watertight curation is applied here as well).

Good ads are relevant.

On amazon affiliates:

The iron rule of nature is: You get what you reward for. If you want ants to come, you put sugar on the floor — Charlie Munger

I don't want to reward the wrong type of behavior.

I don't get your point about affiliate links at all. What better ad is there than a link to a recommended book itself? I'd much rather see you get a small portion of the purchase price of a book you recommended to me than look at your ads, curated or not.

But it's your site! I hope you do find success with the ad strategy, your site seems useful, and I hope it remains worth your time.

This comment puts it well: https://news.ycombinator.com/item?id=17789161.

You have the same incentive to prioritize ads over recommendations. If I trust you to keep ads clearly separated, I also trust you to keep recommendations undiluted.

> You have the same incentive to prioritize ads over recommendations.

No.

> If I trust you to keep ads clearly separated, I also trust you to keep recommendations undiluted.

True. I am the one curator at the moment. I'll open this so that others can curate / recommend. I require that dilution of intention doesn't happen then. DNA is important.

>> No to amazon affiliates.

> Thanks you! This shows character and actual, proper altruistic intentions.

I hate ads and every other scheme to extort money from people, but I have no problem with affiliate links and do not understand why would someone have a problem with them. Could you, or someone else who thinks like this explain what problem you have with affiliate links?

Thanks.

It tosses in the incentive to start recommending things based on how valuable the affiliate link would be,diluting the value of the recommendations

This

As a general rule, I steer away from review sites, and instead I look for professionals using the product. For example I follow some photographers, carpenters, etc, and see what equipment they use for themselves and why they use it. They are not reviewing products, they are using them and posting a list of products for my convenience.

If I decide to buy the same equipment, I see nothing wrong using their affiliate links.

This is where I'd like to get to.

Reference: https://news.ycombinator.com/item?id=17793341.

OT: I checked your profile but couldn't find an email to contact you. It'd be cool if I can ping once a while to get feedback on where the project is heading.

I truly hope it works out well for you. My personal response:

* I don't know what is meant by "good" people; I'd choose a different term.

* My impression is recommendations from "famous" people, which don't interest me personally. Famous people aren't necessarily smart and unless a topic is in their specific domain, they have no more to contribute that the average person. (Personally, I find the idolization of famous people as 'smarter than thou' to be disturbing and unhealthy; they are no better than gits1225 and I'd like to see their recommendations too.)

* I would be very interested in seriously thought out recommendations by experts in their domains. For example, for a desk chair, I'd be interested in an anonymous chair designer's recommendation. For a history book, I'd be interested in an historian's recommendation. I'm not any more interested in a famous coder's recommendation for a chair or a history book than I am in any other coder's recommendation. (To be fair, some of the famous people are experts making recommendations in their domains.)

All valid points which I too am pondering on but don't have clear answers.

I checked your profile but couldn't find an email to contact you. I'd love to discuss details once I have written down my thoughts.

Maybe there would be a chance these experts would answer an email, some of them at least.. the problem would be, of course, to identify them.

I agree. The hack I'm using is PeopleRank (check my comment above).

Thanks for checking my project. Glad HNers loved it. I had a great day.

Thanks to Alexander and Michiel for pointing me to Knuth's and Keith's recommendations.

I'm signing off for today. I'll reply to new comments (if any) tomorrow.

Have a great day : )

I suggest adding zebra striping to the list. It's hard to follow some of the short titles over to the names.

I see. Hmm. Thanks for the feedback.

Great idea. Could you make the links to Amazon referrals so you can make some cash at the same time? https://affiliate-program.amazon.com/help/node/topic/2021023...

I appreciate the tip and I did consider it, but chose not to.

In that case maybe make them smile.amazon.com links so some proceeds go to charity?

That is a lovely idea. Let me sit on the suggestion.

I tried this for http://wanderkind.org/ and IMO it ended up being more than helpful.

Affiliation only seems worth it if you have a lot of traffic. Even then, maybe not--unless you're making serious money, I'd rather build genuine trust.

This is a passion project for me so making money is not the priority.

I can appreciate that, and it is a refreshing attitude.

I greatly appreciate this attitude, and now I want to donate. Do you have a donation option?

I take tips to good recommendations at aj@abyjames.com : )

I'm not exactly sure how it'd be implemented with the current design, but I'd think it'd be useful to have a query in the URL so you could link http://abyjames.com/projects/recommendations?query=Books to just show books, rather than pointing someone to http://abyjames.com/projects/recommendations and asking them to type in "Books".

Yea. I built the basic idea. I'll be building features one by one. Thanks for the feedback.

- Someone compiled a list called "Andreessen's Favorite Books": https://docs.google.com/spreadsheets/d/1mmEKySwi0iYjWx_DoigL...

- I compiled a list (see comment on video) from a trailer video for John Doerr's "Measure What Matters": https://www.youtube.com/watch?v=8aW5gdRRn_U

I was going to add Andreessen's recommendations this week. This helps!

Thanks for pulling the book recommendations. I'll add 'em this week as well.

Might I suggest not using the urls for search? Searching for "Software" gives you a ton of Joel Spolsky recommendations because of his website URL ;)

Yea. This is a wart. I'll be fixing it.

Love this idea. I've thought about doing something similar, but as a graph, so we could see books that many people have recommended in common.

I'm working on it. Would love to get feedback once I build the feature. I checked your profile but couldn't find an email to contact you.

Awesome, thanks! dave@davebryand.com

Email saved. Thanks.

Hack: You can search for a book (The Mythical Man-Month) and see who all recommended it.

Thanks so much for implementing this idea. Very nice idea. Whenever i want to watch a movie its hard to find whats best with imdb or google search because its so generalized. But if a movie is liked by John Carmack, Michael Nielsen, Steve Yegee or Elon Musk then the chance of me liking is really great.

Also the citing the source might be a good place to use block chain, just a thought.

That's the idea. Glad you liked it : )

Nice!

I have a request though - could you make it more obvious that clicking on the name of the recommender takes you to their review/the context of why they recommended the item?

That wasn't obvious to me at all, but it's often really useful to understand why the person recommended the item in the first place, and what you should expect to get out of it...

Thanks!

Thanks for the feedback. I'll see how I can make it obvious.

I made it obvious. Let me know if this is clear: http://abyjames.com/projects/recommendations/.

Thanks! Much better, but you might want to display it a bit differently to avoid having the word `source` repeated constantly...

My suggestion would be to replace `source` (apart from the title) with some kind of icon for example.

Good luck!

Hmm. Yea. Thanks for the feedback.

Love this, cheers!

For some reason, the design of this is very sweetly nostalgic to me--reminds me of the web from 10y ago (in a good way!).

Web fonts and accessibility concerns with small font size have made good ol 13px Helvetica with blue links on white bg rare.

The design is deliberate. I love the old web too : )

One suggestion: - Order by column (especially by date for tech).

Thanks for the feedback. It is on my to build list. Bumped it up : )

Love the site. I see the "Recommendation" header center-aligned, while all other headers are left-aligned. It looks odd to me, was this intentional?

It is. Left aligning that header makes it look weirder. This looks better (IMO).

When I searched for 'Software', all of Joel Spolsky's recommendations showed up because the word 'Software' is in his source URL.

Yea. This is a wart. I'll be fixing it.

One tiny suggestion (well, medium-sized :)

Often, these recommendations come with a reason why this is a good book. You might want to include them, if possible.

It is (but I see it is not obvious). Click on the person's name.

I changed the header name from person to source. Does that make it obvious?

Somewhat more. Maybe change it to "Source article" so people know there's actual text behind that?

Clicking on a name, I'd expect either a bio, or a list of things they recommended. It didn't occur to me it might be the original recommendation. (But then, it might just be lack of coffee :)

Either way, thank you for building this and collecting the data!

I made it obvious. Let me know if this is clear: http://abyjames.com/projects/recommendations/.

Awesome, thank you!

This is a great start!

How do you account for duplicates? i.e. Seveneves

The idea (posted elsewhere in this forum) of linking them to smile.amazon.com might be a great suggestion.

That is a feature (in my book). Search for The Mythical Man-Month tells me that the book is recommended by Marc Benioff, Joel Spolsky and Jeff Atwood. That is valuable information for me and tells me that I should read the book as soon as possible : )

I agree on smile.amazon.com.

> That is a feature

Agreed

I miss filter by category (I have tons of books on TODO list already and I wanted to quickly check everything beside books).

You can search in tags (example: Philosophy, Book), but it doesn't do what you are looking for.

Thanks for the feedback. I'll look into it.

Also put a drop down menu because I don’t know if my query in the search box is going to exist.

really cool, i especially like the product (ie non-book) recommendations. having a curated list of products isnt something i'd seen before, while there are plenty of reading lists. your book list is really great but for me i was most pleasantly surprised seeing the non-books

Glad you found it useful : )

Great idea! How do you collect data?

This data is from my bookmarks.

First thought was to search for Billy G, but it seems you haven't included him yet.

These recommendations are from people I follow, and I'll be gradually adding people one recommendation a time.

I couldn't find relevant information on Google for Billy G. Do you mind sharing his profile?

It's Bill Gates

Ah! I follow his book recommendations and I'll be adding them this month.

This is a great idea. I recommend using ArangoDB ( graph database ) to store the data. You can then support interesting features very easily.

Is there a way that the PageRank algorithm could be applied here?

I'm gonna use a different hack: PeopleRank. Example: If Paul Graham recommends a person, then I'll add all their recommendations.

It is 1 or 0 : )

You get one of those two scores for scalability.

Really cool. How long did this take to build?

1 night to build. 1 week to curate content.

Great. I've been working on something similar but in Arabic. Thank you for this list.

You are welcome : )

Care to share your work?

Nice, here goes my weekend.

I recommend Norvig's recommendations : )

Thanks, very useful!

You are welcome : )

not enough fiction recommendations :)

Haha. True. That should change once I add recommendations from people from different backgrounds. Maria Popova is not going to recommend The Mythical Man-Month : )

Reply on news.ycombinator.com