I often don’t agree with your conclusions, but I find your data-heavy writing style compelling and always gives me something to think about. I also appreciate how you engage with your readers, and even those like me who have questions. So, I’ll keep on reading and learning and sometimes arguing here. Thanks for writing.
I saw the value in your writing immediately and became a paid subscriber. I see the likes for your entries are never as high as they should be but am happy to hear there are many who do read you. What’s next? I do send my MP links and hope he reads them. I hope others do and we may eventually be able to move this beast of government to a better position. Good luck in your second year!
Keep up the good work. Your work points out the failures of "idealogical-driven"policy. Data-driven policy should be the goal of all levels of government, no matter what their stripe.
Not sure how I originally stumbled into this substack but I became a free subscriber instantly and recently upped to paid status. Looking forward to reading more in 2025.
I ask for paid upgrades because it could free up enough of my time from my normal professional responsibilities to devote more energy to new Audit-related projects (I have a very long list of topics and tools I'd like to address). But I would never want to pressure any subscribers to push themselves beyond their comfort zone.
In the meantime, I make a decent living in the technology world, which lets me publish most of my new posts freely for everyone. And I have no plans to change that.
I got attracted to your posts because you make innovative use of data, as opposed to spin rhetoric. I'm not sure I like and trust the work of your AI assistants, however. I'd like to have more insight in how much you curate and validate their output. Do you think you could enlighten us about that? In general terms would be good but even better would be to tell us what you did specifically for any given AI-driven post. Thanks! I look forward to your posts in 2025.
That's a good question. As a rule, I treat stuff that comes from AI pretty much the same way I treat other resources like Statistics Canada data, academic papers, and media reports. In other words, I assume there's probably some level of bias or factual problems, so I'll do as much fact checking as I can, but I present the results to my readers with full transparency: this is what I've seen and this is where I saw it.
The (partial) exception is my Parliamentary Briefers where, for all intents and purposes, the summary service AI gives me can't be duplicated by human journalists for any money. I've been told by government officials that they're working hard to use very similar AI tools to reproduce what I'm doing for internal consumption: it's literally the only way for all the text-based data that government produces to ever be utilized.
In the Briefings posts, that's largely correct, yes. The difference is that I generally don't cross-validate Statistics Canada data or academic publications even for the regular posts, but I will, as a rule, fact-check AI output in the regular posts.
I have been pondering your query, "Where to from here?" for a few hours now. I don't have a good answer but I have my best answer. In other words, is my best good enough? Dubious, to be sure, but it is my best.
But first, but first.... congratulations on your anniversary; I am certain that your wife is thrilled - oops, wrong anniversary. I have been trying to think about when I became a subscriber - yes, paid - and it seems to me that it was somewhere about a year ago but I was not one of your original three and perhaps my advanced age and my enjoyment of your work has caused / permitted delusions as to the time I have been a subscriber.
Where to from here? As I said above, this is by best answer but is very likely not at all good enough. Anyway, anyway ....
In thinking about where to it is important to first consider where from and, in that context, I would argue that you start from a base of excellent work. I find your efforts astounding from the perspective of, of course!, your topics, the methods that you have devised to develop analyses and your initial conclusions. I specify "initial" conclusions as you offer some obvious conclusions arising from your data but you are careful to not hit the reader over the head with those conclusions and you allow we readers to think about and draw further conclusions, including contrary conclusions. I am intrigued that you are finding ways to lightly use AI.
Okay, that is all preamble and I have not yet addressed the question but, as I noted, to try to determine where to it is necessary to know where from.
It seems to me that the quality of your work is clear and that if you had further subscribers (paid, to be sure!) you could afford to engage assistants, whether AI generated or contract workers on particular projects. The problem is, that is in the realm of fantasy right now.
Therefore, I suggest that you do something terrifically boring and tedious: keep doing what you are doing in terms of your quality work. I further suggest that you might consider a series of topics that might be somewhat provocative (whatever they may be; I don't know, but provocative). Further, further, you might think on those provocative topics and think of some other relatively successful Canadian substacks [my particular favorite is The Line with Gurney and Gerson] and talk to them about some of your provocative analyses and see if there is any cross fertilization that might occur. I know that I read substacks and it is not unusual for there to be comments about yet other substacks, quotes from them, etc., not in the sense of advertising but recognition of quality work.
To carry on my thought about provocation / provocative. Perhaps you might query readers about potential projects, together with whatever basic information a reader might possess. You do not want to get into partisan efforts but providing analytical analyses as you have previously done is a very good start. Any partisan can themselves take data points and skew it as they wish but developing useful analyses for discussion is not partisan. At least, that is my contention.
Again, I don't really have an good answer for your question but I have provided my best response. And, again, again, congratulations on your anniversary.
Thanks for this. On that cross-fertilization note, my collaboration with Macdonald Laurier Inst. was an attempt in that direction - and it did bring some success.
Oh, and our (38th) anniversary is actually just two days hence.
So, David, you can have a minor celebration of The Audit's anniversary but it had best be a major blowout for the 38th. [I will be at 53rd anniversary next year so I do know about blowouts.]
I often don’t agree with your conclusions, but I find your data-heavy writing style compelling and always gives me something to think about. I also appreciate how you engage with your readers, and even those like me who have questions. So, I’ll keep on reading and learning and sometimes arguing here. Thanks for writing.
I saw the value in your writing immediately and became a paid subscriber. I see the likes for your entries are never as high as they should be but am happy to hear there are many who do read you. What’s next? I do send my MP links and hope he reads them. I hope others do and we may eventually be able to move this beast of government to a better position. Good luck in your second year!
Good luck and success in your next year.
Keep up the good work. Your work points out the failures of "idealogical-driven"policy. Data-driven policy should be the goal of all levels of government, no matter what their stripe.
Not sure how I originally stumbled into this substack but I became a free subscriber instantly and recently upped to paid status. Looking forward to reading more in 2025.
Keep on digging! Thank you for keeping on digging.
I have a few (too many?) paid subscriptions but am still freeloading here for now. I did buy your book, though!
All the best for the next year and beyond. If my pension increases by 2% or more I’ll sign up for a paid subscription.
I ask for paid upgrades because it could free up enough of my time from my normal professional responsibilities to devote more energy to new Audit-related projects (I have a very long list of topics and tools I'd like to address). But I would never want to pressure any subscribers to push themselves beyond their comfort zone.
In the meantime, I make a decent living in the technology world, which lets me publish most of my new posts freely for everyone. And I have no plans to change that.
I got attracted to your posts because you make innovative use of data, as opposed to spin rhetoric. I'm not sure I like and trust the work of your AI assistants, however. I'd like to have more insight in how much you curate and validate their output. Do you think you could enlighten us about that? In general terms would be good but even better would be to tell us what you did specifically for any given AI-driven post. Thanks! I look forward to your posts in 2025.
That's a good question. As a rule, I treat stuff that comes from AI pretty much the same way I treat other resources like Statistics Canada data, academic papers, and media reports. In other words, I assume there's probably some level of bias or factual problems, so I'll do as much fact checking as I can, but I present the results to my readers with full transparency: this is what I've seen and this is where I saw it.
The (partial) exception is my Parliamentary Briefers where, for all intents and purposes, the summary service AI gives me can't be duplicated by human journalists for any money. I've been told by government officials that they're working hard to use very similar AI tools to reproduce what I'm doing for internal consumption: it's literally the only way for all the text-based data that government produces to ever be utilized.
Must I understand then that, apart from asking the question (which you do show) you take the output as is, without any cross-validation or editing?
In the Briefings posts, that's largely correct, yes. The difference is that I generally don't cross-validate Statistics Canada data or academic publications even for the regular posts, but I will, as a rule, fact-check AI output in the regular posts.
I have been pondering your query, "Where to from here?" for a few hours now. I don't have a good answer but I have my best answer. In other words, is my best good enough? Dubious, to be sure, but it is my best.
But first, but first.... congratulations on your anniversary; I am certain that your wife is thrilled - oops, wrong anniversary. I have been trying to think about when I became a subscriber - yes, paid - and it seems to me that it was somewhere about a year ago but I was not one of your original three and perhaps my advanced age and my enjoyment of your work has caused / permitted delusions as to the time I have been a subscriber.
Where to from here? As I said above, this is by best answer but is very likely not at all good enough. Anyway, anyway ....
In thinking about where to it is important to first consider where from and, in that context, I would argue that you start from a base of excellent work. I find your efforts astounding from the perspective of, of course!, your topics, the methods that you have devised to develop analyses and your initial conclusions. I specify "initial" conclusions as you offer some obvious conclusions arising from your data but you are careful to not hit the reader over the head with those conclusions and you allow we readers to think about and draw further conclusions, including contrary conclusions. I am intrigued that you are finding ways to lightly use AI.
Okay, that is all preamble and I have not yet addressed the question but, as I noted, to try to determine where to it is necessary to know where from.
It seems to me that the quality of your work is clear and that if you had further subscribers (paid, to be sure!) you could afford to engage assistants, whether AI generated or contract workers on particular projects. The problem is, that is in the realm of fantasy right now.
Therefore, I suggest that you do something terrifically boring and tedious: keep doing what you are doing in terms of your quality work. I further suggest that you might consider a series of topics that might be somewhat provocative (whatever they may be; I don't know, but provocative). Further, further, you might think on those provocative topics and think of some other relatively successful Canadian substacks [my particular favorite is The Line with Gurney and Gerson] and talk to them about some of your provocative analyses and see if there is any cross fertilization that might occur. I know that I read substacks and it is not unusual for there to be comments about yet other substacks, quotes from them, etc., not in the sense of advertising but recognition of quality work.
To carry on my thought about provocation / provocative. Perhaps you might query readers about potential projects, together with whatever basic information a reader might possess. You do not want to get into partisan efforts but providing analytical analyses as you have previously done is a very good start. Any partisan can themselves take data points and skew it as they wish but developing useful analyses for discussion is not partisan. At least, that is my contention.
Again, I don't really have an good answer for your question but I have provided my best response. And, again, again, congratulations on your anniversary.
Thanks for this. On that cross-fertilization note, my collaboration with Macdonald Laurier Inst. was an attempt in that direction - and it did bring some success.
Oh, and our (38th) anniversary is actually just two days hence.
So, David, you can have a minor celebration of The Audit's anniversary but it had best be a major blowout for the 38th. [I will be at 53rd anniversary next year so I do know about blowouts.]