There is a particular kind of chaos that happens when two people are trying to choose the same thing. Choosing something for yourself is already a miniature negotiation between your budget, your expectations, your questionable optimism, and the part of your brain that suddenly decides a feature you learned about eleven minutes ago is now absolutely essential. Add another human being and the entire decision grows another head.
Maybe you are buying a new TV together. One person wants the best picture quality possible. The other is staring directly at the price tag wondering when televisions became small business investments. One person cares deeply about gaming performance. The other wants something that looks good during movies and does not require a doctorate in menu navigation. Neither person is wrong, and that is the annoying part. A lot of shared decisions are not really disagreements. They are two perfectly reasonable sets of priorities arriving at the same table with different spreadsheets.
That is exactly the sort of problem I wanted Octivary's Choice Compasses to handle better. So I recently added the ability to open a secondary Choice Compass that links directly to the first one. Two people can now configure their own priorities separately, and the scoring system takes both sets of preferences into consideration before sorting the listings. Naturally, this has also opened an absolutely delightful mathematical can of worms.
The Second Choice Compass Has Entered the Building
Choice Compasses already allow someone to work through different criteria, select what matters to them, and arrange those sections by priority. Listings are then scored and sorted according to those choices so that the most relevant options rise toward the top.
The important part has always been that the results are being ranked for the person using the compass, not according to whatever product happens to be popular this week, somebody else's "Top 10 Things You Absolutely Need" article, or whichever company bought the loudest marketing megaphone. The scoring is meant to answer a much more useful question: which of these options fits what I actually care about?
Eventually there was a very obvious problem sitting in front of me. What happens when "I" becomes "we"?
Big decisions especially tend to involve more than one person. Cars, appliances, houses, vacations, furniture, electronics and even something that seems fairly harmless, like choosing a coffee maker, can unexpectedly become a summit meeting involving counter space, milk frothers, cleaning difficulty, pod ecosystems, water reservoirs, coffee strength, and one person insisting that they will definitely use the iced-coffee setting every morning. Sure. Absolutely. We all believe you.
The collaborative feature gives each person their own Choice Compass instead of forcing two people's priorities into the same control panel.
[ User 1 priorities ] + [ User 2 priorities ] = hopefully less chaos
One person can configure their compass, close it back up, and hand the proverbial controls over to User 2. The second person can then make their own selections without staring at everything User 1 picked first. That part is surprisingly important to me because it creates a little pocket of independence inside a shared decision.
Two People Can Choose Without Accidentally Influencing Each Other
When two people are sitting beside each other filling out the same list of requirements, there is an unavoidable bit of social gravity. One person clicks something, the other notices, and suddenly the second person's choices may start drifting toward the first person's answers. "Oh, you picked that? Yeah, I guess I care about that too." But do you, or did you just watch somebody else click it?
Collaborative Choice Compasses can currently support a maximum of two users, and one of the things I really like about the design is that the first person can enter their configuration and collapse their compass before the second person begins. User 2 does not have to build their preferences while staring at User 1's answers.
That means there is less opportunity for someone to unconsciously adjust their preferences based on what the other person selected. Both people get to gather their thoughts first, then we bring the data together.
For certain decisions, that could make the result much more meaningful. If two people are choosing a home, for example, one might independently prioritize commute distance, garage space and price while the other puts schools, yard size and kitchen layout at the top. Nobody has to start the conversation by sacrificing what they actually want. The overlap can emerge from the scoring instead.
User 2 Is Not the Sidekick
This has also created one of the most important pieces of testing around the feature: the two users have to remain equal. User 1 might care about completely different things than User 2, and they might arrange their priorities in completely different orders, but User 2's opinion cannot become mathematically weaker simply because they happen to be using the secondary compass.
That would defeat the entire purpose. If User 1 says Feature A is extremely important and User 2 says Feature B is extremely important, the system has to respect both of those positions when calculating the final ranking. User 2 is not Robin, User 1 is not Batman, and nobody gets the Batmobile because they opened the page first.
Both sets of priorities have equal standing in the overall calculation. That does not mean every individual choice inside each compass has identical weight. The entire point of a Choice Compass is that higher-priority sections should influence the score more than lower-priority ones. But the people themselves remain equal participants.
That distinction has become one of the major things I am testing. If the collaborative ranking says Listing A is first, I want to know it got there because the combined scoring genuinely supports it, not because User 1 quietly received a bigger mathematical microphone.
Yes, It Is in Production. I Still Consider It Testing.
The collaborative feature is now in production, but I still very much consider it to be in testing. Part of the reason is that I am wandering into territory that is new to me.
Weighted scoring itself is certainly not some shiny new invention. Humans have been creating structured ways to compare options for a very long time, and formal decision-analysis methods have decades of history behind them. What is new to my eyes is what happens when I start applying that thinking to two independently configured Choice Compasses that need to remain equal while still respecting the internal priority structure of each person.
From everything I have personally seen while wandering around the internet, I have not come across this exact approach implemented quite this way. Maybe somebody has built something similar in a dusty academic corner somewhere. Maybe there is a spreadsheet from 1997 doing it magnificently. The internet is enormous and I am not about to plant a flag and declare myself Supreme Inventor of Two People Having Opinions.
But I have not personally found another consumer-facing filtering system behaving quite like this, which means I keep finding new little mathy rabbit holes. And apparently I live here now.
When Does the Second-Best Listing Become the Best Listing?
One of those rabbit holes involves sorting. Suppose both users heavily prioritize a certain feature. Great. A listing has that feature and climbs the rankings. Easy enough.
But what if the products that satisfy the highest-priority choices perform terribly everywhere else? What if another listing misses that one ideal feature but strongly satisfies six other important preferences? At what point should that other listing take over the top spot?
That question becomes even more interesting when there are two users involved. Maybe the number-one listing satisfies User 1 beautifully but does a mediocre job for User 2. Meanwhile, Listing B does not completely dominate either person's compass but performs consistently well across both. Should B win? Sometimes, probably yes. But where exactly should that crossover happen?
That is the kind of thing I get to ponder about now. I have been testing different sorting approaches and looking closely at the point where unavailable high-priority options should stop overpowering everything beneath them.
The goal is not to create rankings that stubbornly worship the highest-weighted criterion no matter what. That would be a terrible decision engine. Imagine saying, "Yes, this refrigerator lacks enough storage, does not fit the available space, costs $1,800 more than planned and has reliability concerns. HOWEVER, both of you wanted an ice dispenser." Winner.
Absolutely not.
The scoring needs enough flexibility to understand the whole picture while still respecting priority. That is where things become far more interesting than simply assigning points and adding them together.
The Selection Details Are Doing Heavy Lifting
Another thing I did not want was for the collaborative feature to produce a shared score and simply tell everyone to trust it. Scores without explanations make me itchy.
Listing A: 91. Listing B: 84. Listing C: 79. Wonderful. Why? What happened? Which preferences mattered? Did one person's requirements get flattened somewhere inside the calculation?
The collaborative results continue showing selection details from both users. That means people can inspect how a listing performed against each person's choices instead of staring at a mysterious number dropped from the heavens. I want the ranking to start a better conversation, not replace the conversation.
You might discover that the number-one option satisfies nearly every important requirement for both people. Fantastic. Or you might discover that the second-ranked option loses a couple of points on something relatively minor for you but significantly improves one of the other person's major priorities.
Now there is something useful to discuss, which is considerably better than standing in an appliance store under lighting normally reserved for interrogation rooms saying, "I don't know. I just like this one more."
Fun Fact: This Whole Thing Started With Wedding Dresses
There is actually a strange little origin story buried underneath all of this. The very first Choice Compass demo I ever made was for wedding dresses.
This was far enough back that the concept of AI helping you write code was basically, "Interesting idea. Good luck out there," and off you went into the wilderness.
At the time, I imagined myself eventually presenting this scoring concept in a room full of executives. In my head, the room would probably have been majority men with a smaller number of women. So naturally I thought: wedding dresses.
I figured the men in the room might pay less attention to the product itself and more attention to how the scoring logic was behaving. I don't know. That was genuinely part of my thought process. Maybe the women in the room would get a little smile out of watching a room full of executives analyze neckline styles, silhouettes and trains with extreme corporate seriousness.
"Please turn to slide 14. We need to discuss the scoring implications of cathedral-length lace."
Perfect.
That demo never became the destination, obviously. I ventured far beyond wedding dresses, but I still find it funny that this entire system, which is now expanding into collaborative weighted decision-making, started with me thinking wedding dresses might be the perfect distraction while I demonstrated scoring logic.
Where Collaborative Choice Compasses Could Really Matter
Naturally, adding this feature caused my brain to immediately start brainstorming markets where it could be genuinely useful. The first place my mind goes is real estate.
Buying a home together is practically the final boss of collaborative decision-making. Price, location, commute, schools, property taxes, lot size, bedrooms, garage, basement, renovations, transit, neighbourhood, distance from family. Somebody wants acreage. Somebody else has absolutely no intention of spending Saturday afternoon operating a lawn tractor. A shared Choice Compass suddenly starts making an awful lot of sense.
Vehicles are another obvious one. Stay tuned. 😉
Cars involve a gigantic pile of competing priorities, especially when more than one person will use the vehicle. Fuel economy, cargo space, reliability, safety technology, seating, performance, price, drivetrain, towing capacity, comfort and approximately six billion trim configurations are all standing nearby waiting to complicate your afternoon.
Large appliances are another category where I think collaborative filtering could have real value. A refrigerator, washer, dryer, range or dishwasher can cost serious money. Choose the wrong one based purely on internet vibes and you may be living with that decision for years.
The consequences are considerably more expensive than ordering the wrong phone case. In the 2026+ economy, big-ticket purchases deserve more thought, especially when two people are involved. It would be nice to have a clearer direction before several thousand dollars leaves the building.
The Rough Part Right Now
There is another side to building something unusual, and it is significantly less fun. Octivary takes in many different kinds of products, services and experiences, then compiles information into structured Choice Compasses and allows users to evaluate those options according to their own priorities.
That makes perfect sense to me. Apparently, it may make considerably less sense to search engines.
Because the site does not fit neatly into one traditional cookie-cutter category, I suspect search systems may have a harder time understanding what Octivary actually is and why its pages are useful. The result right now is that the site can feel like it has been pushed onto a remote island somewhere around page 847. Fantastic weather. No visitors.
I find the situation deeply ironic because the long-term goal is to build something that could become extremely useful for choosing things across a wide range of markets. But if a search engine sees an unusual structure, many categories, structured product data and scoring tools and decides, "I have no idea what this creature is," that usefulness becomes much harder for people to discover.
I am still working through that, and I am very open to feedback from people who understand search, structured content, discovery or simply happen to notice something Octivary could improve. If there is a better way to communicate what these pages are, how they are different, and why the Choice Compasses provide value beyond ordinary product lists, I want to hear it.
I would very much like to be rescued from Search Engine Castaway Island, but I also cannot let that stop development. There is no point sanding every interesting corner off a project just to force it into a shape that already exists everywhere else.
In Brighter News, Octivary Is Starting to Knock on Doors
While I continue tinkering with scoring logic, ranking behaviour, collaborative filtering and the great search-engine mystery box, Octivary has also started reaching out to markets that could potentially benefit from what we are building.
It is much too early to know what will come from those conversations. Maybe something, maybe nothing, maybe someone reads the email while eating lunch and thinks, "What on Earth is a Choice Compass?"
That is part of building. Eventually you have to stop polishing the machine in the garage and roll it outside where other humans can look at it.
The encouraging part is that the collaborative feature makes the larger concept considerably easier for me to imagine in real-world situations. I can see how this could fit into real estate, vehicles, appliances, travel, household subscriptions, large purchases, and anything where one person's "must have" is another person's "I genuinely do not care about that."
That is the territory I want to explore.
Going Beyond the Ordinary Filter
I keep calling these things filters because that is the easiest word to understand, but I increasingly think Choice Compasses are becoming something beyond a normal filter.
Traditional filters are binary little creatures. Under $1,000? Yes. Has Bluetooth? Yes. Four-wheel drive? No. Remove it. Useful, absolutely, but weighted decision-making asks something different.
How well does this option fit? How important were the things it matched? What did it miss? What happens when another person has a completely different set of priorities? And can we combine those priorities without quietly giving one person more influence than the other?
That final question is now one of the most interesting parts of Octivary for me. The collaborative feature is only supporting two people right now, and that feels like the correct place to start.
Two humans, two independently configured compasses, equal influence, one ranked collection of options, and plenty of mathematics lurking underneath trying to behave itself.
The goal is still not to make the decision for anyone. It is to take a giant pile of choices, gather two people's thoughts separately, bring them together fairly, and make the resulting chaos a little easier to understand.
Because if two people are about to spend thousands of dollars on something, "we saw good reviews and got a vibe" probably should not be the entire decision-making framework.
A clearer direction would be nice, and if Octivary can help provide that direction while also preventing one couple from spending three Saturdays debating refrigerators, I will happily put that one in the win column.