That's an interesting phrase. This should really be implicit in everything we say or do. It stands to reason, if something doesn't work, you try something different.
But when you're famous and have thousands of people who constantly try to undermine you or misquote you or take something out of context, you start to need to be explicit in calling this out. Otherwise next month's tech rag headline says "YC partner recants failed investment policy".
I think the startup-specific part is being willing to identify mistakes.
My version of this problem is almost always having to insert the word "generally" into generally almost all of the things that I state (note the redundancy in that sentence).
A corollary to this is to have to preface things that you say in order to not offend one group or another or a particular person that was held in high esteem by others.
"But when you're famous and have thousands of people "
Same issue that celebrities have. We can call it the "sucks to be you" problem. You become so famous that you can't even go out without getting accosted by photographers.
I'm sure it only took a couple rounds before people started noticing you (and some alums) in the cap tables and realizing what's going on. Word spread. I have no doubt that now it is considered a signal.
Investing is like dating. You need to be right for them, and they need to be right for you. Just because you don't want to invest in someone doesn't mean they're a bad investment. Smarter angels & VCs probably realize this, but there's plenty of perfectly good money that would not go to companies who might be a good fit amongst the less savvy.
That said, quietly using partner/alumni investment as a signal for VCs doesn't quite fit the "Startups are a perfect meritocracy" narrative I'd prefer to believe...
Interestingly, searching for "Matt Maroon" (on Google, Crunchbase, AngelList, etc) yields results like  that only mention Blue Frog. Same for your blog. Wouldn't care otherwise, but I wanted to know if your story could be true or to dismiss as an internet rant.
However there was one interview-like-thing  still hanging around the Google Index that connected DraftMix  with a "Matthew Maroon" as the CEO 
People claiming to have gone though YC on HN without a known name will get called out. Hang around HN long enough and you'll know who is who.
quietly using partner/alumni investment as a signal for VCs doesn't quite fit the "Startups are a perfect meritocracy" narrative I'd prefer to believe...
What led you believe that? I'd guess startup investment is 90% social, 5% metrics and 5% technical. Witness, for example, the YC investment thesis of it being all about the team applying. Team dynamics aren't a meritocracy (except in the most broad sense I guess).
I agree with the other commenter. You spent way too much time spying on Matt Maroon today.
I'm sure the YC Partners' track record of "picking" is no better than SV Angel, a16z, Greylock, etc.
I think this tweak to the investment policy will contribute to eliminating the sub-par angel investors who base a majority of their investments on which companies received funding from YC partners. PG taking action to make his dreams a reality 
I don't think this is the case; they only need to infer "better than average" or "not a lemmon". The problem is everyone else is trading-off the research of others and abnormal volatility# results, which has dis-utility to the companies involved. In fact, it eventually would become a dis-incentive to join YC. It is this latter outcome that surely YC wants to avoid. They are not going to benefit from the pricing of current companies, so much as they would lose long-term value of their franchise. Ie, if it becomes a very "hit or miss" investment of Founder's time, the more capable founders may seek to mitigate their exposure to such a process.
# Not bias, per-se
However, the downside of not ranking is that your top 1-5% of students are not clear to the extreme elite (Harvard, Stanford, MIT) schools, meaning there is no one who is "guaranteed" those positions, as even your top 5% has to compete with students in the top 20 or 30% for admittance. For high schools that couldn't hope to admit more than 5% of their class in top tier unis, it makes sense to rank (as then you get to showcase your top students as being exceptional). If you expect 20% or more to be competitive candidates though, it's usually better to stay quiet on relative success.
So not inflation, just that the student body was limited to kids that were going to make high grades anyway with very heavy AP course loads (which push GPAs over 4.0 since they're weighted).
The only reason it's bad is because there are lots of great startups in the batch that can be hurt by it, which is obviously not what the partners intend.
Investors "treat investment by YC partners as an accurate sign of how promising we thought a startup was." So then, the company raises more money - and appears to look more promising.
Sure, partners may have less incentive immediately (as mathattack said) - but the real winners will be more clear to the partners after demo day (and three weeks).
They will have a list of winners (in their mind) before demo day. Then, investors at demo day will pick a list of winners (who they invest in). And YC partners will see the overlap - those who are picked as winners by YC Partners and independent investors.
This policy affects investments made by individual partners. I'm sure 500 startups does not stop it's mentors from investing right after demo day, or does it explicitly do that?
P.S: I'm the co-founder of a company who was in Batch 6 of 500startups.
I also point this out because what I've found in business is that so many basic principles end up repeating themselves in different situations. Things that you don't learn in school or in books but by listening to others and real life experiences. One reason that certain people who have grown up in business families have a nice advantage over those that don't.  You have a seat of the pants feel for things because everything is just a variation of something that has already happened.
 But it's not just growing up in the family but also interacting and listening and thinking. My siblings grew up with the same parents but are vastly different than I am in their thinking and understanding even given the same trough of water.
Hmm. The way to effect this as a broker is to take comission, with 3rd part capital backing the acquisition. The broker "pump and dumps" the property and takes a comission. The structure is opaque to the purchaser, and avoids the signalling 'tell' alluded to in the narrative above. This isn n% as profitable as a outright "front-run" or "flip", but can be executed at greater leverage/scale, since it is easily repeatable and the conflict of interest is far less obvious.
Since YC itself doesn't do follow-on investments it doesn't have to worry about competing with its own partners (unlike the VC LPs).
This new policy seems to be addressing a specific problem: demo day is meant to be an opportunity for all the presenting startups, not a competition for attention between them. I.e. for YC's sake, it should never hurt a startup to present at demo day, so its a problem if not having a yc partner invest looks like a signal of relative weakness.
The main concern is this: let's say your fund passes on Twitter during its seed round, but you put $25k into it personally. 8 years later, the company has an IPO and you get a 100x or 1000x return. LPs are going to be frustrated that they let you invest on their behalf, but you took such a great investment for yourself. Even if the fund passed, there will always be the perception of "maybe they just passed so that Partner X could invest individually because the company looked very promising."
I never knew fund raising worked so quickly.
That kind of money would usually be put in by a larger fund. Most angels write checks for $25k-$100k.
* One difference being that the demo day investors will have preferred shares upon conversion and YC takes the same common shares that the founders and employees get.
My sense is net this is positive.
If it is trustworthy, then surely this would mean that great startups prosper faster and those that aren't that great know it even earlier, meaning that they can pivot after spending much less time on building their business.
If it is not trustworthy, the numbers will show that to be the case, with investors being able to directly assess which companies backed by YC-partners actually make it big. In this case, surely investors will wise up and not go by YC-partner investments as a signal of success.
If I'm missing something, I'd appreciate it if someone could point out where my logic falls short. Otherwise, I stand by my conclusion that this seems unnecessary.
This "peace of mind" factor has significant economic value, in that it frees up resources from "monitoring" potentially ambiguous motives and situations, and this improves the throughput for tasks related to running the business and making profits.
With respect to the "market equilibrium" notion, theory and practice differ. Gaming the (any) system increases in liklihood as the un-eveness of outcomes is exaggerated; information flows become more assymetric; and/or they become more opaque.
That is precisely the description of early stage investments. More or less. Hopefully this makes a bit of sense.
The counter-argument as to "letting the market" reveal the information, is that it is unlikely to do so. There simply are not formal mechanisms for privately trades companies to disclose such information in a timely and transparent way. Remember, the companies are private explicitly to prevent this from occurring.
Aside from that, if PG feels that the signal is overvalued, the remainder of his own portfolio becomes toxic unnecessarily. So it's a bad outcome if the signal is accurate, and a worse outcome if it isn't.
Early investment by any of the YC partners after a startup has been accepted to YC could give the impression that the YC partners may be doing some sort of insider trading based on the meetings that they had with the company. That probably isn't an issue at all, but the perception is there.
I think I might just be missing the point entirely. We are not talking about publicly listed companies, so if you're smart enough to have fashioned a position as a YC-partner for yourself, then kudos to you and surely that just challenges external investors to get better at picking YC winners even earlier.
An analogy to this is what happened during the financial crisis when they got all banks to agree to take money so as not to send a signal showing what banks were the weaker ones.
Think of it as a star which is trying to explode because of all those gases burning away, and gravity is keeping it together. The size of the star is the equilibrium point of the differential equations describing this dynamic.
One of the signals that was keeping the 3 week window to 3 weeks could have been the YC investments (gravity), but now that gravity has been set to a lower level, the 3 week window will expand..to account for this.
Curious whether this will actually matter in the long run but I guess they'll adjust if needed.
YC is in a strong enough position that they are likely not dependent upon one particular partner, and most partners would prefer to have access to future batches. They probably don't need to have a contract that slaps on additional legal damages, because at such an early stage the small gain for favoring their own investments does not outweigh the longterm gains from continued early access.
Those startups that join demo day realize that they're putting themselves on a public platform to be assessed, etc. If they don't get funded by YC, that's their problem; not YCs.
There's a problem with PGs policy since it rules out the possibility of YC being a first rounder. Which is a huge reason why VCs exist in the first place.