The website service Myhome.ie has “quoted” prices displayed but it also has “hidden” prices (provided by the vendor) which are not displayed but which determine the ordering of houses when any search is ordered by price by the website user.
To see this in action, sort on all south dublin houses listed at greater than 500,000 (it works most dramatically at the high end where there has been the most uncertainty in price guidelines). After the search contents have appeared, hit the “price” button to sort the search contents from highest to lowest price. The ordering is quite different from the ordering of “quoted” prices! There is a rough decline in quoted prices, but it is choppy, and shows some of the hidden intentions of the vendors in terms of price discounts.
I do not know for sure that this is the algorithm as I have described it. I am only inferring from the treatment of POA (price on application) listings, which also appear at particular (but undisclosed) price-points on a search ordered by price. I have no direct information only supposition. The ordering is definitely a little bit different from that based on quoted prices, and this is not a “computer bug” in the search code since the price ordering continues to make sense in a general way even if it is not “monotonic” in the quoted price.
I have my doubts. I think it’s more likely that the properties are ranked by date when they have the same price. It’s also more likely that they are ranked by price/sqft than there being a ‘hidden’ price.
@Coles2 — It is not just a secondary ranking (eg, when two prices are the same, rank by date or square footage) it is an over-ride ranking which supersedes quoted price. The website specifies that it will rank by price when the user hits that button, but then it ranks by a hidden price or other attribute not the quoted price. In many cases the displayed ordering does not match the ordering of quoted prices (particularly for properties near the south Dublin top end) rather than just choosing how to relatively rank same-price properties. More than that I do not know.
It doesn’t stand up to scrutiny when you test in from a software testing perspective.
It only happens when price sorted asc (rather than default desc).
If you look at an area, the apparent price difference for a house that appears out of order is much greater than if you look at many areas together (eg. look at 4 beds in Sutton then look at 4 beds in Dublin North).
In most scenarios house is one place off (unless multiple houses off).
Out of order houses appear to have an ID > 200,000
This is a simple software bug.
My guess is they are using a sharded database (multiple databases) and they’re inserting certain houses before rather than after the house they should be inserted beside in the list.
You may be correct about sharded databases – I have no expertise in that. However please note it works whether one uses ascending or descending so you need to check that again. I just did it with the case I mentioned and it is definitely mis-ordered in both cases, but only for the top-end properties as far as I can tell. It may be as you suggest a computer bug - but not one within my limited knowledge of likely bugs. Where is the alternative information used for sorting coming from? It is not random - it keeps a semblance of ordering, but inexact.
Maybe its sorting according to price range, ie, 500k-600k rather than differentiating between 560k and 585k. That way everything gets lumped in together within a certain range. Perhaps after that the order is to do with time on the market, number of bedrooms etc.