After being away from GD expired auctions for some time, I decided to try again. A few scenarios observed during the test (short time test, may not be significative) are below.
I would not list exact domains, as HugeDomains may be following this thread, no need to give them extra tips. Due to the same reason, most of amounts mentioned are not 100% exact amounts occured. Of course initial $12 bid is still $12, no changes here.
First scenario
*most typical case* - Candidates for closeouts, which were converted to auctions because of one initial $12 bid, placed in the last minutes.
Hugedomains joins as the second bidder almost instantly. In all cases, hugedomains used above-mentioned proxy bid (9-12% of the GD value). Making them pay smth. around this amount (or at least 8%-10% of the GD value) is possible then. I personally did this a few times, and they never disappointed me. They always won. In a couple of cases, I was trying to decide whether to let them have the domain for $100 (their current bid) or to force them to pay $200 (as GD value was $2000+). Even before I decided, somebody else (who looked and smelled like a bot) promptly joined and increased the hugedomains winning bid to $200. Fascinating. It was so elegant. I'd prefer to think that another bot was not looking to win the domain, but tried to make hugedomains pay as much as possible instead. Side effect: mentioned non-hugedomains bot became 2nd highest bidder (placed $195 bid and hugedomains outbid him with $200). I used bold font in the previous phrase, will later explain why.
Extra observation No 1: A half of domains I played with here (and placed the 1st $12 bid) were just... too average. They do have some sense, but normally I would not bother even to handreg such stuff, saying nothing about bidding. Moreover, I selected these "average" domains after my inhouse online+offline listfiltering script excluded them. In other words, to the best of my knowledge, there was nothing in these domains could allow an automated script to mark them as valuable for resale. So, I guess that hugedomains did not mark them for grabbing in closeouts, and, therefore, they joined just because somebody else showed interest. GD value for all domains in question was $1000+, so it is unclear would they autojoin auctions with GD value of 3 figures or less.
Extra observation No 2: Due to an average quality of domains I played with in this part, I did not try to outbid hugedomains. However, as noticed earlier in this thread, hugedomains would frequently stop and the human will win paying ~10-13% of the GD value. If so, hugedomains will become the 2nd highest bidder. Again, I used bold font in the previous phrase, and will later explain why.
Second scenario
*rare case, saw it two times only*
Also - candidates for closeouts, which were converted to auctions because of one initial bid, placed in the last minutes. However, hugedomains joined as the 3rd or 4th bidder. Due to unknown reasons, he used a proxy of just ~5%-6% instead. Which is why, for the purposes of analyzing test results, I categorized this case as rare.
One domain: many more bidders joined, and one of them won paying enduser price after some "war". Hugedomains was outbid very fast, and did not place any extra bids.
Another domain: two bots joined (or humans behaving like bots). I'd call them "BotAlpha" and "BotBeta". It looked like BotAlpha thought that BotBeta is hugedomains, and vice versa. Because each of them tried to let another one pay as much as possible, up to *9-12% of the GD value* which, as we know, is an usual hugedomains proxy. One of them won the auction, possibly inadvertently (as real hugedomains was using ~5%-6% proxy).
Third scenario
*Auctions with high activity (a lot of bidders, a few days actions length)*
Cannot say a lot here. Even if I liked the domain (not all cases), I was not willing to pay even the current highest bid, so I ignored almost all such auctions. I participated in maybe 2 or 3 auctions and placed just 1 bid (where the "current price vs quality" combination was somewhat acceptable). Naturally, I was outbid. Naturally, hugedomains was inside. He was also outbid, but his behavior was very interesting. He joined in last minutes (not a surprise), but, instead of his "normal" proxy (9-12% of the GD value) he applied the following formula: current high bid + $10 ($15, 20) - just to cover 2-4 extra steps. It is also curious that this proxy was in fact ~6% of the GD value, so he knowingly elected to pay less than he is usually ready to. And, he did this with domains of better quality. Why? Hmmm... maybe he joined just to use a chance to become the 2nd highest bidder?
Fourth,... scenario(s)
The test was limited, so there must be other HD-related scenarios I missed.
Closeouts
As expected. Anything valuable is grabbed by bots, most notably hugedomains. GoDaddy changed the closeout appearance time from random to strict (+ 5 sec. or so). How convinient for bots. GD claimed "equal access for all", but the reality is very different. As noticed a few pages before, hugedomains is now shown with bidder id 913933 and earlier it was 91932. Two bots? It would mean 2x chances to grab a closeout. Maybe hugedomains has 100 GD accounts and 100 bots, exclusively for closeouts?
The last but not the least.
What I marked in bold above (a few times). It looks and seems that on many occasions bots (hugedomains and others) are not actively trying to win, but are trying to become 2nd highest bidder instead. What if the price paid (the highest bid) is shared with the second-highest bidder, for selected customers if they are the 2nd? Under NDA. An extra incentive. Nothing illegal here, it also happens offline. Sometimes it is public, somewhat similar occured with ICANN nGTLD strings recently auctioned.
In other words, what if you outbid hugedomains (proxy $200 on 2K appraised domain) with $205, you pay $205 to GoDaddy, and GoDaddy pays revshare to hugedomains because they were the 2nd? It is just a conspiracy theory. No proof so far. And, in any case, it is not a wrongdoing from strictly legal point of view...