Oh snap! I went to some random PG essay to see which ad agency PG would put up and was immediately disappointed to see an ad for Mortgages, which was entirely off-base for the page, but then an up/down mod appeared! What?! That was awesome! Down clicking made a 'Sorry' box appear and the ad was gone! Huzzah!

But still, what will happen when an advertisement that a group pays for is consistently down-modded into oblivion, essentially nullifying any chance for it to appear? How will the YC startup handle the irate company? I know an answer would be that they should change their advertisements, but still, some companies may get angry at the YC startup that the money they payed for advertisement didn't do any good. How will they handle that situation? I'm guessing the advertisements are charged on a per-click bases, but is there a base charge, or tiered pricing for size of initial random viewing of web-pages registered with the company?

If the YC startup can't fulfill a promise of at least some level of ad-viewer-ship, then why should a company have their ads go through them?

Maybe the YC company only charges advertisers for each impression -- or maybe for each click-thru. If an advertiser's goods are "down-modded into oblivion", then the advertiser at least didn't have to pay anything. And they can take their ad to, say, AdSense, or some other less exclusive network.

So, what you're saying is "Who would want to be rewarded based on performance, given that some people won't perform well?" I guess I might be irate if I said "Let my customers tell me how good my ads are," and the answer was "They suck," but wouldn't I be better off knowing, or at least having the option to know?

Yes, a reasonable person would be glad to have such feedback, but I was more curious of the Marketing Management reaction who would look at their ad click-through's and notice that they aren't that high for a service they just paid money for, or that they're ads aren't even being viewed anywhere at all.

I was just asking how an online advertising service, especially a small startup, handle such things.

But what about the marketing manager's manager. If the marketing guy says "We could use this new ad service -- but we'll only get clicks if users like our ads, and that's not going to work out well for us," he'd deserve to lose his job.

Nobody who says "the more accurately you judge my efforts, the worse off I'll be" is worth keeping around.

I was trying to spark conversation along a different line of dialog, but maybe I have been misunderstood. Yes, this new ad system is a great idea - instant feedback and better, more specific targeting for individual websites and viewers, and any ad/marketing group should love it. Yet a certain amount of unpredictability remains, especially for ad click-through rates, and it is this that I wanted to follow.

Right now, it is relatively easy for a company to sign up for an online advertising service with some minimal guaranteed ad views. From that companies can project future earnings based on their click through and purchase rates, even if their ads are terrible, and nobody except for 0.1% click, they at least have a base line for money coming in that is fairly reliable. This system, however bad for the end users experience, is extremely helpful for the businesses planning and forecasting. Now, the CFO can say, 'We're going to have at least this much money!', and Net Ops groups can say, 'Well that means we'll probably need this much bandwidth!', and the entire business can some-what get along.

Then, Adpinion steps in. Suddenly, Advertising/Marketing can't predict how much, or even if, they'll be able to meet previous projections, projections are useless at this point. Now the CFO can't say if the company will be able to meet market predictions, and the Net Ops guys will be in the dark about if they'll suddenly be crushed by users' bandwidth requests or that they've over allocated bandwidth and they're now stuck with some fat pipes that are just getting spam messages. Well, this is all a little dire, but the point is that Adpinion can create contentions in a business.

Ah! But here is what I wanted to follow through on - even if Adpinion can create unpredictability in long running advertising strategies, the real win here would be for a one-two punch with traditional online advertising venues. A business could market test half their advertising revenue, find what works and then send it through traditional online ad streams. Now that business can up their average click rate by quickly market testing ads, and they can make their advertising team hungry and on-edge about what how well that day/week/months' ad scheme performed. And this is where Adpinion wins, for me. From an overall business perspective, outside of a localized ad/marketing department, it would provide such a benefit to any company that they would be senseless not to use it.

Leon, yeah! We're excited about the potential for our Adpinion as a tool for businesses to quickly evaluate the performance of their ads.

The issue of advertisements getting shown less because everyone dislikes them is something we're looking at closely. We focus a lot on individual user preferences, so if you dislike an ad hopefully there is someone out there who likes it and they will get shown the ad more to keep the number of impressions consistent. However, there will be cases where an ad gets less impressions because the majority of people dislike it. We feel that serving these ads as much as the ads that everyone likes creates an inefficiency in the cost per click model: if you're a website publisher and there are two ads that pay the same per click, you're not making as much off the less popular ad. We're exploring ways to incorporate ad ratings into the cost structure in a way that gives advertisers incentive to make well-liked ads.

Also: guaranteeing a minimum viewership/maximum price for cost estimation is definitely something we're aware of, and we have some ideas on how to achieve that. We'll reveal them when we're done ; ).

One thing I will give the YC guys though is that you can make some hell of user friendly website. I know what the site does within 5 seconds. Anyone has a few spare hours to spend on my website?

It's an exciting idea because it has that magic "Why didn't anyone do this before" feel to it. That's always a good sign. Good luck. I haven't had any coffee yet but I totally missed the voting thumbs on the left, and I was looking for them. Not sure if that's good or bad though.

There was a lot of hype when Seth Godin implemented this idea on Squidoo back in December. I haven't heard much about it since. That's not to say it's a bad idea, but rather that it's not some miracle technology with guaranteed easy money.

I thought it had a lot of potential at the time and I still think it has a lot of potential now. The only problem is that until there is a large body of ads you get very scatterbrained results.

There's an automatic large body of ads because their default ad is the Amazon affiliate program. While there's not a lot of money to be made as an Amazon affiliate, there's useful information to be gained about a user's taste in media.

I hate ads. Everyone hates ads. Why would we take what is largely an automated process (extracting information from a site to guide ads), and make humans do it?

With the idea to "make something people want", isn't it a bad idea to make something a minority want and the vast majority will fight against? Especially considering the majority's clicks are all that matter.

When they say they will make something people want, they mean the advertisers. Besides, this is the Internet. Ads are here to stay, so why not make them suck less than Google's Adsense?

My point is that advertisers will like this, content creators might like this if they get more money, but the USERS of the site will not like it. I'm not saying this is worse than other image ads, though it is presumptuous to try to waste my time in the process of trying to sell me something.

"Ads are here to stay"

This isn't true. Ads needn't dominate the way people make money on the web.

They certainly suck more than text ads.

Hallo Ivan! The issue for us is that the automated process in place Does Not Work. For you, if you don't interact with the system at all, we just treat the site you're at as an aggregate "user" and guess what you'll like that way. We think that's better than just guessing based on the site content. -Luke

Good luck!

I don't hate ads. I hate companies that sell products or services that only benefit themselves. When ads come from legitimate businesses providing products/services that are actually valuable, I think it's a wonderful thing that it's getting a lot easier to reach people who need them.

I hate ads. I ignore most of them, but distracting animated ads get adblocked. If adpinion gives ad-haters an easier way to make an ad disappear, adblock will get used less and they'll get more information.

(Even though I hate ads I think they have their place.)

Hm, interesting, but slightly weird concept. If the users are willing to vote actively for their favorite ads, why not let them do it on your homepage to begin with?

On the other hand, why voting, isn't click/no click sufficient information to judge the effectiveness of ads (seems to work for Google Adwords)?

That's a really good point about the click/no-click. That said, I could see myself down-voting ads for which there's no chance I'd ever click, leaving ads I might be interested in at some point, but didn't actually want to click at that moment.

Also, there's the type of ad where you think you might be interested in it, so you click it to see what it's about, only to find that the product or whatever actually sucks. That's where the down-vote would help.

I don't know if this idea will work, though. Assuming it's used by web surfers, then I think advertisers will want it. But that's a big assumption. What's the incentive for web surfers to use it? Looking at ads, even "high-quality" ones, isn't really a priority of mine. And there are other tools like StumbleUpon or del.icio.us that suggest content to me, but don't really have the commercial agenda that an ad does.

I wonder when people are going to get it. We have been talking about "IN-vertising" for ever but NO, NO, NO. We still want to guess what the user wants. No software can tell what a user wants better than the user him/herself. I believe we should spend more time and resources providing users with services that let the user initially decide the ads they want. For me send anything that rhymes with coffee, mobility, startups and thick women. Block the other ads. If I feel like buying a car in the next few weeks, then let me pull the firefox or IE plugin and tell it to serve me a few VW Golf ads and listings. Would it be hard to build such a plugin?

If you already know you want a VW Golf, the ad isn't very useful for the advertiser.

Yes it is. If there are 4 cars dealers in my state who have what I am looking for, they can all serve ads to me, then I compare and buy the best car with the best payment option.

I may be a minority, but that's not how I buy stuff if I know what I want to buy. If I know what I want to buy, then I research the item using Google.

I don't click on ads often, but when I do, it's usually because something caught my attention that I wasn't directly searching for.

The one exception is that I occasionally try to "trigger" ads that give me a discount once I know exactly what I want.

I see. I assumed the advertiser was VW, but as you point out, there are others who'd still be interested.

That's probably when the ad is the most useful, because they can get you to buy it sooner. It's a pretty good deal for VW if a few cents' worth of internet ads can cause you to make a $20k purchase a month sooner.

http://www.kuro5hin.org/ had ads that you can comment on etc. but it never worked out. These new ads are a step forward with the targeting system.

Still, I'd like to know how many people actually care about seeing ads.

Reddit too, also failed there as far as I can tell: http://ads.reddit.com/

This doesn't mean it wouldn't work across a large set of sites though. Aggregating responses and correlating them should produce really effective targeting.

This is not such a great idea, but it probably can be tweaked.

First, it needs some automated targeting. Does anyone know how AdSense targets ads? I bet the ad ends up on the pages that would be returned from a Google query containing the keywords. The better pagerank, the more expensive the ad. I do not have in mind a good algorithm for adpinion, but they should figure something out - Google's targeting is really good..

Second, upclicking doesn't matter and the button is not even needed. The only way the user indicates he likes the ad should be to click through.

Third, do not just count votes. Implement the best open recommendation algorithm from Netflix prize contest instead.

Downclicking that hides the ad is good and useful.

Its plastered all over paulgraham.com, not just the essays. I personally like the ad free version of paulgraham.com :). I wonder if adpinion is another YC startup.

Yep. Why else would I do it?

To make tens upon tens of dollars!

It's all about the Hamiltons.

Selling out for kicks and giggles?

PS: Reddit for ads sounds clever, but still won't get me to click them.

To give proceeds to News.YC readers? :-)

I got a chuckle that you had to link to Adpinion, because practically every page on paulgraham.com has already been submitted.

So what kind of information can the adpinion guys extract if someone downvotes every single ad?

Hi ph0rque, the theory is that even the most ardent of ad-downvoters would eventually like an ad enough to not down it (or, that they would get bored of compulsive negativity, and start being more selective). That one ad would be an extremely valuable piece of information. -Luke

Sorry, but that doesn't make sense. The haters won't down-vote. They just won't click at all.

That's still a piece of information though, especially if the same user has actively down-voted other items.

There is no useful information gained if people never touch the system. Clearly non-downvoting means something for people who periodically downvote. But the way I plan to use the service is ... not at all.

I see, I miss-read what you wrote. I had thought we were still talking about "ardent down-voters" Yes, in order for them to glean any information from a particular user there will have to be some clicks. And indeed, the most common way that a user uses this system might very well be "not at all".

Hi Luke, are you with Adpinion? Can I advertise for http://www.getkratom.com on Adpinion?

Contact Luke at lukexi@gmail.com

Get rid of "any" kind of ad from that page coz its annoying the hell out of visitor(s) :).

Actually that's what the system would do. Why expend bandwidth annoying people?

Well, we know what we have to do, now that you've given away the secret :~).

This will be an interesting experiment to watch. Its taking the reddit/digg style voting to the world of internet ads and hoping that users will tell what kind of ads they want to see. If every user is like me, it'll mean no ads... maybe everyone is not like me.

Actually, most people aren't. Apparently, most people actually LIKE ads--so long as they are relevant.

I don't know about you guys, but I actually enjoy looking through the ads in Wired. Usually they're catchy and they are almost always relevant for me. I can't say that about most ads on the web, and certainly not for TV (I will never buy a truck in my life so stop 'effin showing me ads for them!) Obviously the problem with these last two mediums is that they don't have enough information about me.

Honestly, I would submit my personal tastes/preference to some sort of central advertising database which authorized companies could then plug into. Right now, it would make a lot of sense if Google hosted that database, but it should really be something open so that any advertiser can use it.

Edit: so long as they are relevant and done well. Despite, for example, my desire to buy a Wii, I would not tolerate a "hit the monkey and win a Wii" ad.

I would suggest instead of "Sorry" say, "OK, Bye." I think the latter would be more satisfying.

I adblock only distracting animated ads. Will they notice a pattern if I downvote according to ad style, or is it only based on ad topic?

I like "Sorry" actually. It may help raise awareness of how annoying and intrusive bad ads are. "Ok, bye" would be appropriate if it was a friend of yours showing you the ad rather than a faceless corporation.

I hate ads as much as the next guy. But somehow, I still voted up an ad at the end of a pg essay. It worked on me.

The UI is good, the idea is really elegant, and it is well executed.

I did the same thing, but I'll bet you did it for the same reason as me -- you wanted to try it out because it was a YC company and you were curious.

Now I wasn't "lying" when I up-voted the ad -- it did seem generally interesting (some Ruby site that provided tutorial screencasts), but I'm not sure I would have voted if it wasn't linked off news.yc.

You're right. I was well-disposed because I figured it was a YC startup.

So how do you get a sample set that's not biased? What kind of benefit is there to the user in saying that the ad worked the way it should?

The user who doesn't like an ad benefits from thumbs down because the ad disappears and he's less likely to see similar ads in the future.

The user who likes an ad benefits from thumbs up because he's more likely to see similar ads in the future.

Hi Paul, another Paul here. We're interested to know if we can integrate Adpinion with AsiaPay.Com maybe you can email me at paulpajo [at] gmail